AVERTISSEMENT

Ce document est le fruit d'un long travail approuvé par le jury de soutenance et mis à disposition de l'ensemble de la communauté universitaire élargie. Il est soumis à la propriété intellectuelle de l'auteur. Ceci implique une obligation de citation et de référencement lors de l’utilisation de ce document. D'autre part, toute contrefaçon, plagiat, reproduction illicite encourt une poursuite pénale. Contact : ddoc-theses-contact@univ-lorraine.fr

LIENS Code de la Propriété Intellectuelle. articles L 122. 4 Code de la Propriété Intellectuelle. articles L 335.2- L 335.10 http://www.cfcopies.com/V2/leg/leg_droi.php http://www.culture.gouv.fr/culture/infos-pratiques/droits/protection.htm

LASERS A CAVITE VERTICAL EMETTANT PAR LA

SURFACE DANS L’ULTRAVIOLET PROFOND A BASE

DES MATERIAUX BAlGaN

BAlGaN-BASED VERTICAL CAVITYSURFACE-EMITTING LASERS OPERATING IN DEEP UV

REGION

A DissertationPresented to

The Academic Faculty

By

LI Xin

In Partial Fulfillmentof the Requirements for the Degree

Doctor of Philosophyin

Physics

Universite de LorraineEcole doctorale: EMMA

December 2015

Copyright© 2015 by LI Xin

LASERS A CAVITE VERTICAL EMETTANT PAR LA

SURFACE DANS L’ULTRAVIOLET PROFOND A BASE

DES MATERIAUX BAlGaN

BAlGaN-BASED VERTICAL CAVITYSURFACE-EMITTING LASERS OPERATING IN DEEP UV

REGION

Approved by:

Advisor: Pr. Abdallah OugazzadenSchool of Electrical and Computer Engieering,Georgia Institute of Technology

Reporter: Pr. Ferdinand ScholzInstitut fur Optoelektronik, Universitat Ulm,Germany

Reporter: Pr. Francois H. JulienInstitut d’Electronique Fondamentale, Univer-sity Paris-Sud XI

Co-advisor: Dr. Frederic GentyLaboratoire Materiaux Optiques, Photoniqueet Systemes (LMOPS), CentraleSupelec

Examiner: Pr. Bernard GilLaboratoire Charles Coulomb (L2C), Univer-site Montpellier 2

Examiner: Dr. Sophie BouchouleLaboratoire de Photonique et de Nanostruc-tures (LPN)

Examiner: Pr. Jean-Paul SalvestriniLaboratoire Materiaux Optiques, Photoniqueet Systemes (LMOPS), Universite de Lorraine

Date Approved: December 2015

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi

ACKNOWLEDGMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

RESUME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

CHAPTER 1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . 91.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.2 Research problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121.3 State-of-the-art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.4 Scope of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

CHAPTER 2 FUNDAMENTALS OF MATERIALS AND EXPERIMENTS . . 232.1 Fundamental properties of III nitrides . . . . . . . . . . . . . . . . . . . . 23

2.1.1 Structural properties of III nitrides . . . . . . . . . . . . . . . . . 232.1.2 Optical properties of III nitrides . . . . . . . . . . . . . . . . . . 25

2.2 Experimental methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302.2.1 Metal-organic vapor-phase epitaxy . . . . . . . . . . . . . . . . . 302.2.2 Characterization techniques . . . . . . . . . . . . . . . . . . . . . 35

CHAPTER 3 MOVPE STUDIES OF BAlGaN MATERIALS . . . . . . . . . . 423.1 MOVPE growth of AlGaN single layers . . . . . . . . . . . . . . . . . . 43

3.1.1 Control of composition and relaxation . . . . . . . . . . . . . . . 433.1.2 Critical thickness for AlGaN grown on AlN templates . . . . . . . 463.1.3 Estimation of threading dislocation densities by XRD . . . . . . . 51

3.2 MOVPE growth of BAlN with high boron content . . . . . . . . . . . . . 543.2.1 BAlN/AlN grown at 1000 ◦C . . . . . . . . . . . . . . . . . . . . 543.2.2 BAlN grown at low temperature with annealing . . . . . . . . . . 60

CHAPTER 4 DEEP UV AlGaN MQWS: DESIGN, GROWTH AND CHARAC-TERIZATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

4.1 AlGaN MQW design for enhanced TE (E⊥c) emission . . . . . . . . . . 704.1.1 Principles of AlGaN band structure calculation . . . . . . . . . . 704.1.2 Design of AlGaN MQWs . . . . . . . . . . . . . . . . . . . . . . 73

4.2 Growth and characterizations of MQWs . . . . . . . . . . . . . . . . . . 764.2.1 Structural characterizations . . . . . . . . . . . . . . . . . . . . . 764.2.2 Optical characterizations . . . . . . . . . . . . . . . . . . . . . . 80

4.3 10- and 20-period MQWs . . . . . . . . . . . . . . . . . . . . . . . . . . 824.4 Defects in MQWs and their influence on DUV emission . . . . . . . . . . 84

iii

4.4.1 Structural investigations of defects . . . . . . . . . . . . . . . . . 884.4.2 Optical influence of defects in AlGaN MQWs . . . . . . . . . . . 92

CHAPTER 5 DISTRIBUTED BRAGG REFLECTOR: SIMULATIONS ANDREALIZATION . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

5.1 Transfer-matrix simulations of DBRs . . . . . . . . . . . . . . . . . . . . 985.2 Design of BAlN/AlGaN DBRs . . . . . . . . . . . . . . . . . . . . . . . 1015.3 Simulation of structural quality factors . . . . . . . . . . . . . . . . . . . 103

5.3.1 Roughness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035.3.2 Influence of lattice strain . . . . . . . . . . . . . . . . . . . . . . 106

5.4 Realization of BAlN/Al(Ga)N DBRs for DUV . . . . . . . . . . . . . . . 1095.4.1 Growth conditions for DBRs . . . . . . . . . . . . . . . . . . . . 1095.4.2 BAlN/Al(Ga)N DBRs with reflection at DUV wavelengths . . . . 1125.4.3 Characterizations of DBRs and reflectance comparison with sim-

ulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

CHAPTER 6 CONCLUSION AND PERSPECTIVE . . . . . . . . . . . . . . . 1236.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1236.2 Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1266.3 Publications and awards . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

iv

LIST OF TABLES

Table 1 State-of-the-art distributed Bragg mirrors based on AlGaInN for wave-lengths below 300 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Table 2 Properties of III-nitrides binaries [1]. . . . . . . . . . . . . . . . . . . . 25

Table 3 Bandgap bowing parameters of III-nitrides ternaries. . . . . . . . . . . . 26

Table 4 Polarization elastic parameters from the literature. . . . . . . . . . . . . 28

Table 5 Properties of metal-organic precursors of III elements. . . . . . . . . . . 35

Table 6 Parameters used for the estimation of critical thickness. . . . . . . . . . . 51

Table 7 Threading dislocation densities of AlN templates and AlGaN layers de-termined by XRD rocking curves (FWHM determination and linear fit-ting lead to an estimated error of 15%). . . . . . . . . . . . . . . . . . . 53

Table 8 Estimated defect densities (FWHM determination and linear fitting leadto an estimated error of 15% for threading dislocations) for sample #1and sample #2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

Table 9 Parameters used of III-nitrides for Eq. 59 . . . . . . . . . . . . . . . . . 108

v

LIST OF FIGURES

Figure 1 Applications of UV light. . . . . . . . . . . . . . . . . . . . . . . . . . 10

Figure 2 Simple schematic of semiconductor LEDs and laser diodes. . . . . . . . 11

Figure 3 Schematic of optically-pumped VCSEL. . . . . . . . . . . . . . . . . . 11

Figure 4 Bandgap energy versus in-plane lattice parameter diagram for III-nitrides. 15

Figure 5 External quantum efficiency as function of wavelength reported at theInternational Workshop on Nitride Semiconductors in 2012 and ICNS2013 [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Figure 6 State-of-the-art optically-pumped AlGaN DUV lasers. . . . . . . . . . . 17

Figure 7 DBR reflectivity for various nitride systems. . . . . . . . . . . . . . . . 20

Figure 8 Wurtzite crystal structure of III nitrides [3]. . . . . . . . . . . . . . . . . 24

Figure 9 Typical crystal orientations and planes of wurtzite III-nitrides. . . . . . . 24

Figure 10 Band alignment of GaN/AlGaN structure [4]. . . . . . . . . . . . . . . . 27

Figure 11 Spontaneous polarization field (Psp) and piezoelectric polarization field(Ppz) for GaN, AlGaN and InGaN coherently strained to GaN (0 0 0 1) [5]. 29

Figure 12 Schematic of III-nitrides epitaxial growth. . . . . . . . . . . . . . . . . . 31

Figure 13 MOVPE system and T-shape reactor chamber. . . . . . . . . . . . . . . 33

Figure 14 Aixtron 3×2 inch, close coupled showerhead (CCS) MOVPE system. . . 33

Figure 15 Al composition of AlGaN layers fully-strained on AlN templates versusTMAl/III ratio. The inset shows the growth rate versus total flow rate of(TMAl+TMG). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Figure 16 (0 0 0 2) 2θ − ω scan with simulation and (1 1 -2 4) reciprocal spacemapping for (a): 24-nm thick AlGaN fully strained on AlN template and(b): 350-nm AlGaN layer with 55% relaxation on AlN template. Bothsamples are grown under a fixed TMAl/III ratio of 57%. . . . . . . . . . 44

Figure 17 Al content in the AlGaN single layers plotted as a function of (a) layerthickness and (b) corresponding layer relaxation for three different TMAl/ (TMAl+TMG) ratios. . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Figure 18 Elongation of a grown-in, threading dislocation to form a length LL’ ofmisfit dislocation line. [6] . . . . . . . . . . . . . . . . . . . . . . . . . 47

vi

Figure 19 Schematic illustration of the configuration of real and image misfit dis-locations in a strained heteroepitaxial structure proposed by A. Fischeret al. [7] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Figure 20 Critical thickness of AlGaN layers grown on AlN templates calculatedby different models along with experimental data. . . . . . . . . . . . . . 50

Figure 21 Skew symmetric scans for of AlN template, 29-nm Al0.57Ga0.43N and630-nm Al0.58Ga0.42N. . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

Figure 22 Schematic of precursors feeding sequence for BAlN and AlN. . . . . . . 55

Figure 23 (a) SIMS elemental concentration depth profiles of B and Al for the sam-ple grown on GaN template; (b) Boron content in solid layers calculatedfrom SIMS by using boron implanted AlN sample as reference. . . . . . 56

Figure 24 (a) STEM images (bright field) of 5-period AlN/BAlN heterostructureand columns are clearly observed in the structure; (b) HAADF-STEMimage to show better contrast of BAlN and AlN layers; (c) high magni-fication of the zone where the 1st BAlN layer starts to grow. . . . . . . . 57

Figure 25 HR-XRD 2θ-ω scans of 5-period AlN/BAlN heterostructure grown on(a) GaN template and (b) AlN template. . . . . . . . . . . . . . . . . . . 58

Figure 26 (a) SEM and (b) AFM images of 5-period AlN/BAlN heterostructure(310 nm for total thickness). . . . . . . . . . . . . . . . . . . . . . . . . 59

Figure 27 Cathodoluminescence spectra at 77 K of 5-period AlN/BAlN heterostruc-ture grown on AlN template. . . . . . . . . . . . . . . . . . . . . . . . . 61

Figure 28 Transmission spectrum at room temperature of 5-period AlN/BAlN het-erostructure grown on AlN template. . . . . . . . . . . . . . . . . . . . 61

Figure 29 Schematic of growth procedure. . . . . . . . . . . . . . . . . . . . . . . 62

Figure 30 (a) HR-XRD 2θ-ω scan of 20 nm BAlN layers grown on GaN template at650 under TEB/III=39%; (b) shows the influence of growth temperatureswhich was varied from 650 ◦C to 800 ◦C . . . . . . . . . . . . . . . . . . 63

Figure 31 HR-XRD 2θ-ω scan of 20 nm BAlN layers grown on GaN templates bycontinuous method under different TEB/III ratio. . . . . . . . . . . . . . 64

Figure 32 AFM images of 40 nm BAlN layers grown on GaN templates underTEB/III ratio of 0% and 39%. . . . . . . . . . . . . . . . . . . . . . . . 65

Figure 33 HR-XRD of 70 nm BAlN on (a) GaN template and (b) AlN templateby FME growth (TEB/III=39%). Inset figures show the smoothing anddeconvolution of two peaks. . . . . . . . . . . . . . . . . . . . . . . . . 66

vii

Figure 34 SIMS elemental concentration depth profiles of B and Al for the samplegrown on AlN template; inset shows the boron concentration obtainedby using boron-implanted AlN as calibration sample. . . . . . . . . . . . 67

Figure 35 (a) Cross section STEM image (bright field) of 75 nm thick BAlN layerscontaining 12% boron along the [1 1 -2 0] zone axis. Zone A has latticeoriented along c-axis and Zone B has columnar feature; (b) higher mag-nification image for the top part of the layer; (c) higher magnificationimage for the film/substrate interface. . . . . . . . . . . . . . . . . . . . 68

Figure 36 Cross section High-angle Annular Dark Field Scanning TransmissionMicroscopy (HAADF-STEM) image of 75 nm BAlN layer containing12% boron; inset shows diffraction pattern after Fast Fourier Transform(FFT). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Figure 37 The band-edge energies with and without spin-orbit interaction. . . . . . 71

Figure 38 Band alignment of GaN/AlGaN structure [4]. . . . . . . . . . . . . . . . 72

Figure 39 Relative oscillator strengths for the optical transitions between the va-lence bands (Γ7 (CH) and Γ9 (HH)) and conduction band (CB) in anAlGaN/AlGaN quantum well as a function of the Al composition in thebarriers, with Al content in the well fixed to xAl = 0.37. The corre-sponding strain in the well is also reported in the top axis. Calculationsconsider that the barriers are strain-free and QWs are fully-strained onAlGaN barriers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

Figure 40 Schematic of MQW structure for emission at 280 nm . . . . . . . . . . . 76

Figure 41 (0 0 0 2) 2θ-ω scan for 4 quantum wells grown on a relaxed buffer on AlNtemplate and the simulation of the structure which used values obtainedby XRD, STEM and EDX analyses. The RSM of (1 1 -2 4) reflection isshown in the inset. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Figure 42 (a) Cross-section High-angle Annular Dark Field Scanning Transmis-sion Electron Microscopy (HAADF-STEM) images taken along <1 1-2 0> zone axis for MQWs and buffer layer; (b) High magnification ofHAADF-STEM images; (c) Al composition map obtained from (b). . . . 79

Figure 43 (a) Cathodoluminescence (CL) spectra at 77 K (and at 300 K in the in-set) for two different values of excitation power; (b) Photoluminescence(PL) at 77 K and at 300 K under excitation of 266 nm; (c) transmissionmeasurements and transfer-matrix simulation of MQWs together withabsorption coefficients (αwell, αbarrier, bu f f er used in the simulation). . . . 81

Figure 44 Symmetric 2θ-ω scans and (1 1 -2 4) RSMs of (a) 10-period and (b)20-period MQWs grown on a relaxed AlGaN buffer on AlN templates. . 83

viii

Figure 45 AFM images of 20-period MQWs grown on a relaxed AlGaN buffer onAlN templates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Figure 46 (a) Cross-sectional HAADF-STEM image of 10-period MQWs grownon a relaxed AlGaN buffer; (b) Compositional mapping obtained fromthe STEM image; (c) Al content distribution along the profiles markedin (a). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Figure 47 Comparison of (a) CL emission wavelength, (b) FWHM and (c) emissionintensity at room temperature between 10- and 20-period MQWs grownon a relaxed AlGaN buffer on AlN templates. . . . . . . . . . . . . . . . 86

Figure 48 CL hyperspectral mapping of 20-period MQWs at room temperature. . . 87

Figure 49 2θ-ω scans of (a) sample #1 and (b) sample #2. . . . . . . . . . . . . . . 88

Figure 50 (a) Skew symmetric ω scans and (b) SEM images of sample #1; (c) skewsymmetric ω scans and (d) SEM images of sample #2. . . . . . . . . . . 90

Figure 51 (a) Cross-section HAADF-STEM image of sample #1 showing 10-periodMQWs without dislocations; (b) low-magnification bright field (BF) STEMimage showing origin of defects; (c) plan-view HRTEM image of AlNtemplate on sapphire showing grains; (d) high magnification image ofgrain boundaries; (e) high-magnification BF image on V-shape pits; (f)high-magnification HAADF-STEM image of 4-period MQWs showingsidewall of V-pits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

Figure 52 Transmission spectra at 77 K and macro-PL at 80 K for (a) sample #1and (b) sample #2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Figure 53 PL integrated intensity as function of temperature (with a cw laser anda weak excitation of ∼ 1W/cm2) and Arrhenius fitting for (a) sample #1and (b) sample #2; (c) is the curve of IQE versus temperature based onArrhenius equation fitting. . . . . . . . . . . . . . . . . . . . . . . . . . 95

Figure 54 CL spectra under 7 keV at room temperature for (a) sample #1 and (b)sample #2; (c) is FWHM of QW emission peaks under different electronbeam energies at room temperature. . . . . . . . . . . . . . . . . . . . . 97

Figure 55 Schematic of electromagnetic wave propagation in the DBR structure. . . 99

Figure 56 Experimental refractive index of BAlN versus incident wavelength [8, 9]. 101

Figure 57 AlGaN absorption coefficient as function of the energy for different Alcontent [10]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

Figure 58 Simulated reflectivity of (a) BxAl1−xN/Al0.70Ga0.30N structure and (b)BxAl1−xN/Al0.80Ga0.20N structure. . . . . . . . . . . . . . . . . . . . . . 104

ix

Figure 59 Lattice mismatch of AlGaN/AlN and BAlN/AlN. . . . . . . . . . . . . . 105

Figure 60 Simulated reflectance of 20-period B0.04Al0.96N (33 nm) /Al0.70Ga0.30N(28 nm) DBR with different values of roughness; Inset shows the curveof the central wavelength reflectivity versus the roughness. . . . . . . . . 106

Figure 61 Simulated reflectivity spectra with and without consideration of the strainfor (a) 20-period AlN/AlxGa1−xN DBR and (b) BxAl1−xN/Al0.70Ga0.30NDBR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

Figure 62 Reflector software interface with factors of surface roughness, interfaceroughness and strain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

Figure 63 Cross sectional HAADF-STEM (BF) images of (a) 75 nm BAlN grownat 650 ◦C with annealing at 1000 ◦C by FME method, (b) 5-pair BAlN(32 nm) / AlN (25) nm grown at 1000 ◦C by FME and (c) 18-pair BAlN(32 nm) / AlGaN (24 nm) grown at 1000 ◦C in a continuous way. . . . . 113

Figure 64 SEM images of (a) 30 nm Al0.82Ga0.18N and (b) 5-period AlN (30 nm)/ Al0.82Ga0.18N (30 nm) grown on AlN/sapphire templates; (c) cross-sectional SEM image of the sample in (b). . . . . . . . . . . . . . . . . . 114

Figure 65 18-pair BAlN/Al0.72∼0.76Ga0.28∼0.24N DBRs with different TEB/III ratios(The composition of boron used in the XRD fittings is estimated by anassumption that it has a linear relationship with TEB/III ratio, and it is15% for TEB/III=39% according to EDX measurements). . . . . . . . . 116

Figure 66 (a) 18-pair BAlN (29 nm) / AlN (29 nm) DBRs reflecting at 260 nm; (b)18-pair BAlN (33 nm) / AlN (32 nm) DBRs reflecting at 280 nm; (c)18-pair BAlN (33 nm) / Al0.8Ga0.2N (24 nm) DBRs reflecting at 265 nm. 117

Figure 67 (a) HAADF-STEM (BF) image of 18-pair BAlN/Al0.8Ga0.2N DBRs re-flecting at 265 nm. (b), (c) and (d) are TEM images with different mag-nifications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

Figure 68 Cross-sectional EDX mapping of 18-pair BAlN/AlGaN DBR reflectingat 265 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

Figure 69 EDX profiles of different elements along the growth direction of 18-pairBAlN/AlGaN DBR structure reflecting at 265 nm with (a) lower and (b)higher magnification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

Figure 70 AFM image of 18-pair BAlN/AlGaN DBR reflecting at 265 nm. . . . . . 121

Figure 71 Comparison between experimental reflection spectrum and simulationsof 18-pair BAlN/AlGaN DBR reflecting at 265 nm. . . . . . . . . . . . . 122

Figure 72 (a) SEM image and (b) AFM image of AlN template grown on sapphirein CCS reactor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

x

Figure 73 (a) VCSEL design with dielectric top and bottom mirrors; (b) VCSELdesign with dielectric top mirror and BAlN/AlGaN bottom mirror; (c)inverted VCSEL structure with BAlN/AlGaN bottom mirror grown onthe active region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

xi

To my parents

To my husband

To my future kids

ACKNOWLEDGMENT

xiii

SUMMARY

The context of this thesis falls in the wide range of potential applications of UV light

sources, such as high-density optical storage systems, sterilization and purification, UV

spectroscopy, environmental control system and medical applications. However, the pri-

mary limitation of current UV light applications is the existing UV sources. The conven-

tionally used sources include excimer lasers, Nd: YAG lasers or mercury lamps, but they

are expensive, and have problems in efficiency, reliability and toxicity. In comparison, the

semiconductor light-emitting devices are an ideal choice due to their reliability, compact-

ness, high efficiency, minimum environmental effects and tunable operation wavelengths by

changing the compositions. On the material aspect, III-nitride (BAlGaInN) semiconduc-

tors are promising candidates since they are chemically and physically stable with direct

bandgaps covering from visible to deep UV (DUV) spectrum. On the structure aspect,

vertical-cavity surface-emitting laser (VCSEL) is one of the most attractive configurations

of semiconductor light-emitting devices considering its low threshold and high efficiency,

and the possibility for the integration of 2D arrays and for the wafer-level tests as well.

It constitutes a multiple-quantum-well (MQW) active region sandwiched by a top and a

bottom distributed Bragg reflector (DBR). However, no VCSELs operating below 300 nm

have been reported until now. The major challenges lie in the two main blocks of VCSEL

structure: the emission efficiency of MQWs and the reflectivity of DBRs, which are lim-

ited by the quality of the substrates and epitaxial layers, optical-polarization properties of

the MQW emission, small refractive index contrast of the layers used for DBRs at short

wavelengths, doping difficulties and others.

The objective of this thesis is to address this need by studying metal-organic vapor

phase epitaxy (MOVPE) growth of BAlGaN materials especially the novel BAlN material,

developing AlGaN MQWs with enhanced TE-polarization (E⊥c) emission and exploring

BAlN/AlGaN DBRs, for the future development of optical-pumped VCSELs operating in

1

the DUV region. The target wavelength is 280 nm here since it’s widely used in the UV

applications such as sterilization and water purification.

The thesis is outlined as follows.

Chapter 1 introduces the project background. The motivation including possible ap-

plications is stated. The research problems regarding different aspects are described. The

state-of-the-art of semiconductor light-emitting devices and DBRs for DUV wavelengths

is also summarized in this chapter.

Chapter 2 presents the basic properties of III-nitride materials as well as a brief intro-

duction of the techniques and equipments used for the growth and characterizations.

Chapter 3 goes into the MOVPE growth and studies of AlGaN and BAlN grown on

AlN/sapphire templates. The first task is to have a careful control of the composition and

the strain relaxation of AlGaN single layers to prepare for the MQW realization in the next

step. The relationship between composition and strain state (composition pulling effect) is

discussed. Different models are brought in to predict the range of critical thickness of Al-

GaN grown on AlN. The method to calculate screw/edge threading dislocation density of

AlGaN and AlN layers from XRD measurements is also established. The second task is to

explore novel BAlN material. MOVPE grown BAlN alloys with high boron incorporation

(11∼12%) have been achieved. The characteristics of BAlN with high boron content are

analyzed. In order to improve the crystalline quality of BAlN, a method of low-temperature

growth with annealing is proposed. The influences of growth temperature and TEB con-

centration in the precursors of III elements (TEB/III ratio) on the boron incorporation and

on the crystalline quality are investigated. It is found that low temperature can alleviate

boron-rich phase poisoning during growth under high TEB/III ratio and improve the crys-

tallinity. Very original results have been obtained showing that wurtzite BAlN layers grown

by MOVPE have clear XRD peaks relating to 12% boron. The studies of the growth and

characteristics of BAlN material can advance the application of MOVPE-grown boron al-

loys in bandgap, strain and refractive index engineering.

2

Chapter 4 is addressed to the AlGaN MQWs emitting at 280 nm. Firstly, 4-period

Al0.57Ga0.43N / Al0.38Ga0.62N MQW structure has been grown on a relaxed Al0.58Ga0.42N

buffer on AlN/sapphire template. The compositions of the quantum wells have been opti-

mized so that the strain present in wells is sufficient to enhance TE-polarized emission. The

relaxed AlGaN buffer on AlN/sapphire template serves as pseudo-substrate, and in this way

the barriers are almost strain-free which limits the formation of strain-related defects in the

quantum wells. Transmission measurements confirm a sufficient oscillator strength lead-

ing to a high optical absorption coefficient in the wells. The results represent an important

step towards the development of DUV light sources, especially for surface-emitting LEDs

and lasers. Based on the results of 4-period MQWs, 10- and 20-period MQWs have been

grown and characterized, which will be processed for devices by depositing dielectric DBR

on the top side and the bottom side. Besides, the typical defects including the threading

dislocations and V-pits in the QW samples have been characterized to help to understand

their origin, propagation and influences on the structural and optical properties, which may

lead to further improvement of the performance of DUV devices.

Chapter 5 is devoted to the theoretical simulations and the experimental realization of

BAlN/Al(Ga)N DBRs. The simulations have been established based on transfer-matrix

method in our group before to give the theoretical reflection spectra. In this work, different

quality factors such as surface roughness, interface roughness and strain in the layers have

been introduced into the simulation and their effects on the performance of DBRs have

been analyzed. For the experiments, the growth conditions for BAlGaN DBRs have been

discussed. 70% reflection has been achieved at 260 nm and 280 nm by using 18-pair

BAlN/Al(Ga)N structures. The experimental data have been compared with the simulations

considering quality factors. Although it still needs more efforts to improve structural quality

and surface roughness, the progress achieved can help to develop final DBRs with reflection

more than 90% and to apply novel BAlN/AlGaN DBRs for DUV resonant-cavity LEDs

(RC-LEDs) and VCSELs.

3

In the end, chapter 6 summarizes the results, perspective, publications, communications

and awards obtained in this work. The research perspective includes device processing and

experiment plans in the new Aixtron 3×2 inch close coupled showerhead (CCS) MOVPE

system, the substrate surface temperature of which can reach 1300 ◦C. The temperatures

higher than 1100 ◦C are expected to favor the growth of AlN, AlGaN with more than 70%

Al and BAlN with less than 5% B, which would be applied to improve the quality and

hence the performance of BAlN/AlGaN DBRs.

4

RESUME

Le contexte de cette these se situe dans les nombreuses applications possibles de sources lu-

mineuses UV, tels que les technologies de stockage optique a haute densite, la sterilisation

et la purification, la spectroscopie UV, le controle de l’environnement, et les applications

medicales. Cependant, le principal obstacle pose aux applications UV actuelles est les

sources UV existantes. Les sources conventionnelles sont les lampes a vapeur de mercure,

les lasers a Nd:YAG ou les lasers excimeres, mais ils sont chers et ont des problemes lies a

l’efficacite, a la fiabilite et a la toxicite. Par contre, les dispositifs lumineux a base de semi-

conducteur sont un choix ideal grace a leur fiabilite, a la grande compacite, a l’efficacite

elevee, aux effets minimaux sur l’environnement, et a leur longueur d’onde accordable

par la modification des compositions. Sur l’aspect des materiaux, les III-N (BAlGaInN)

sont les candidats prometteurs car ils sont stables chimiquement et physiquement, et ils

presentent les bandes interdites couvrant le spectre visible a l’UV profond (DUV). Sur

l’aspect des structures, le laser a cavite vertical emettant par la surface (VCSEL) est l’une

des configurations les plus attrayantes pour les dispositifs d’emission lumineuse a base de

semiconducteurs. Il offre des avantages tels que le seuil bas, le haut rendement, et la pos-

sibilite d’integration des reseaux 2D ainsi que la possibilite de faire les tests au niveau de

la plaquette. Il comporte une zone active de multipuits quantiques (MQWs) situee entre

un reflecteur de Bragg distribue (DBR) superieure et un DBR inferieure. Neanmoins, il

n’existe aucun VCSEL qui peut fonctionner en dessous de 300 nm. Des defis importants

concernant deux blocs principaux de la structure de VCSEL sont a optimiser: l’efficacite

des emissions de MQWs et la reflectivite de DBRs, qui sont limitees par la qualite des

substrats et des couches epitaxiales, les proprietes optiques de polarisation des MQWs,

le contraste faible d’indice de refraction pour les couches dans les DBRs a des longueurs

d’onde courtes, les difficultes de dopage, etc.

L’objectif de cette these est de repondre aux defis releves auparavant en etudiant la

5

croissance de materiaux BAlGaN particulierement le nouveau materiau BAlN elabore par

epitaxie en phase vapeur aux organometalliques (MOVPE), en developpant les MQWs

d’AlGaN avec l’augmentation des emissions TE polarisees (E perpendiculaire a laxe c), et

en explorant les DBRs a base de BAlN/AlGaN, en vue du developpement future de VCSEL

a pompage optique fonctionnant dans la gamme de DUV. La longueur d’onde cible ici est

de 280 nm, parce qu’elle est largement utilisee dans les applications de UV telles que la

sterilisation et la purification d’eau.

La these est decrite comme suit.

Chapitre 1 est une introduction donnant le contexte du projet, les applications possibles

et la motivation. Les problemes de recherche sur differents aspects sont decrits. L’etat

de l’art concernant les dispositifs d’emission de la lumiere a semiconducteur ainsi que les

DBRs dans l’ultraviolet profond est resume dans ce chapitre.

Chapitre 2 presente les proprietes basiques des materiaux ainsi qu’une breve introduc-

tion des techniques et des equipements utilises pour la croissance et les caracterisations.

Chapitre 3 porte sur la croissance par MOVPE et sur l’etude d’AlGaN et de BAlN

elabores sur les substrats AlN/saphire. La premiere chose est de controler avec soin la

composition et la relaxation de contrainte de couches AlGaN afin de realiser la structure

de MQWs dans l’etape suivante. La relation entre la composition et la contrainte est dis-

cutee. Differents modeles sont introduits pour prevoir approximativement l’epaisseur cri-

tique d’AlGaN sur AlN. La methode est egalement etabliee pour calculer la densite de

dislocations traversantes (vis ou coin) dans les couches Al(Ga)N a partir des mesures de

XRD. La deuxieme tache est dexplorer le nouveau materiau BAlN. BAlN a haute teneur

en bore (11∼12%) a ete elabore par MOVPE. Les caracteristiques de BAlN sont analysees.

Afin d’ameliorer la qualite cristalline, la methode de croissance a basse temperature avec

le recuit est proposee. Les influences de la temperature de croissance et le concentration

de TEB dans les precurseurs des elements (TEB/III) sont examinees. On trouve que les

basses temperatures peuvent attenuer la contamination de la phase riche en bore pendant la

6

croissance utilisant un TEB/III eleve et ameliorer ainsi la qualite cristalline. Des resultats

originaux ont ete obtenus sur les couches de BAlN wurtzite elaborees par MOVPE, mon-

trant que les spectres de XRD ont des pics claires correspondant a une incorporation de

12% bore. Les etudes sur la croissance et les caracterisations de BAlN peuvent progresser

les perspectives pour les applications de nitrures de bore, ce qui peut conduire a plus de

liberte dans l’ingenierie de bande interdite, de contrainte et d’indice de refraction pour les

sources eventuelles lumineuses a DUV.

Chapitre 4 s’interesse aux MQWs d’AlGaN emettant a 280 nm. Premierement, la struc-

ture de MQWs comprenant 4-periode Al0.57Ga0.43N/Al0.38Ga0.62N a ete elaboree sur le tam-

pon d’Al0.58Ga0.42N a contrainte relaxee depose sur le substrat AlN/saphire. Les compo-

sitions de MQWs sont optimisees de sorte que la contrainte dans les puits est suffisante

pour ameliorer les emissions TE polarisees. Le tampon a contrainte relaxee elabore sur

AlN/saphire serve de pseudo-substrat, et de cette maniere les barrieres sont presque sans

contraintes qui limitent la formation de defauts dans les QWs. Les mesures de transmis-

sion confirment que la force d’oscillateur est suffisante conduisant a un grand coefficient

d’absorption optique dans les puits. Les resultats representent une avancee importante

vers le developpement des sources lumineuses a DUV, notamment pour les LEDs et les

lasers emettant sur la surface. Compte tenu des resultats de 4-periode MQWs, 10- et 20-

periode MQWs ont ete realises et caracterises, et ils seraient utilises pour les dispositifs

en deposant un DBR dielectrique superieur et un DBR dielectrique inferieur. De plus, les

defauts typiques incluant les dislocations traversantes et les V-defauts dans les QWs ont ete

caracterises pour essayer de comprendre leur origine, propagation et influences sur les pro-

prietes structurelles et optiques, cela pourrait conduire a une amelioration supplementaire

de la performance des dispositifs DUV.

Le chapitre 5 est consacre aux simulations theoriques et a la realisation experimentale

de DBRs en BAlN/Al(Ga)N. Les simulations sont etablies par l’utilisation de la methode

7

de matrice de transfert, pour fournir les valeurs theoriques de reflectivite et les concep-

tions structurales pour les experiences. Dans ce travail, differents facteurs de qualite tels

que la rugosite de surface, la rugosite d’interface et la contrainte dans les couches ont

aussi ete introduits dans la simulation, et leurs influences sur la performance de DBRs sont

analysees. Pour les experiences, les conditions de la croissance pour les DBRs ont ete

discutees. 70% de reflexion a ete obtenu a 260 nm et a 280 nm en utilisant 18-periode

BAlN/Al(Ga)N seulement. Les resultats experimentaux ont ete compares avec les simula-

tions. Bien que des efforts doivent encore etre realises pour ameliorer la qualite structurelle

ainsi que la rugosite des surfaces, les progres accomplis contriburent au rapprochement des

DBRs avec reflexion plus de 90% de reflexion et permettent l’application des nouveaux

DBRs BAlN/AlGaN dans les LEDs a cavite resonante (RC-LEDs) et dans les VCSELs.

Enfin, le chapitre 6 resume les resultats, la perspective, les publications, les commu-

nications et les prix obtenus dans ce travail. La perspective de recherche proposee com-

prend le realisation des dispositifs et les plans d’experimentation dans le nouveau reacteur

MOVPE (Aixtron 3×2 CCS) dont la temperature de la surface peut atteindre 1300 ◦C. Les

temperatures superieures a 1100 ◦C devraient favoriser la croissance de AlN, AlGaN con-

tenant plus de 70% d’aluminium et BAlN contenant moins de 5% de bore.

8

CHAPTER 1

INTRODUCTION

1.1 Motivation

The general objective of this work is to make progress for deep UV (DUV) vertical-cavity

surface-emitting lasers (VCSELs) based on BAlGaN material system.

UV and DUV light sources are in high demand in daily life, industrial and research area,

as shown in Fig. 1. The UV light at sufficiently short wavelengths can be used as a disinfec-

tion method (ultraviolet germicidal irradiation, UVGI) to kill or inactivate microorganisms

or pathogens, since it can destroy the nucleic acids in these organisms so that their DNA is

disrupted [11]. Therefore it is useful for sterilization and water/air purification. UV light

can accelerate the decomposition of organic compounds, which provides better solutions

for environmental pollution and plastic recycling. UV radiation is helpful in the treatment

of skin conditions such as psoriasis and vitiligo (UV light therapy) [12]. Photography by

reflected UV radiation is useful for medical, scientific, and forensic investigations, for the

applications of detecting bruising of skin, alterations of documents, or restoration work on

paintings. UV photography can also be used in astronomy to identify the chemical compo-

sitions of the interstellar medium, the temperature and the compositions of stars [13]. The

fluorescence produced by UV illumination can be used in mineralogy, gemology, chemical

sensors, fluorescent labelling, biological detectors and so on [14]. High-power UV light

can be used for a speed curing process by photochemical reaction that instantly cures inks,

adhesives and coatings [15]. UV light can have immediate applications to largely enhance

optical storage density, since recording density in the compact disk is proportional to the

inverse of the wavelength [16]. UV spectroscopy, along with visible and IR spectroscopy, is

routinely used to qualitatively determine the presence of elements and organic compounds

by using absorption or transparency of the sample [17].

However, the primary limitation of current applications are the existing UV and DUV

9

Figure 1: Applications of UV light.

sources. The conventional UV light sources are mercury lamps, Nd: YAG (solid-state)

lasers or excimer lasers (for example ArF and Ar2 lasers). These sources suffer from low

level of performance, low reliability, significant size, and the emission of toxic substances.

Compared with conventional light sources, the semiconductor light-emitting devices are an

ideal choice due to their reliability, compactness and high efficiency. Besides, the wave-

length could be tuned by changing the compositions of the active region. UV semiconduc-

tor light sources are of great technological interest in our daily lives.

Semiconductor light emitting devices have three configurations: light-emitting diodes

(LEDs), laser diodes (LDs) with edge emission, and vertical-cavity surface-emitting lasers

(VCSEL). Compared with other two configurations shown in Fig. 2, VCSELs have vari-

ous advantages. Figure 3 presents the schematic of an optically-pumped VCSEL, which

consists of top distributed Bragg reflector (DBR), active region containing MQWs and bot-

tom DBR mirror. The emission light from VCSEL is perpendicular to the surface, so it

could be integrated into two dimensional arrays. 10000 devices could be integrated on

one wafer to give high output. Circular beam makes it easy to be coupled into the fiber.

10

Figure 2: Simple schematic of semiconductor LEDs and laser diodes.

Figure 3: Schematic of optically-pumped VCSEL.

High reflectivity DBR mirrors enable low threshold and high output. Besides, it could

be tested on the wafer level during processing which greatly decreases the manufacturing

cost. Compared to edge-emitting lasers (EELs), it has low temperature sensitivity due to

its single-longitudinal-mode cavity.

AlGaInN-based wide bandgap semiconductors with chemical and thermal stability have

brought innovative changes in photonic devices, which enable the operating wavelengths

of LEDs and lasers to reach a spectral range spanning from blue to DUV. However, the III-

nitride VCSELs demonstrated so far operate in the wavelengths of visible violet and blue

spectral range, while no efficient VCSELs operating below 300 nm have been reported. The

11

general objective of this thesis is to develop the VCSELs operating below 300 nm based on

the BAlGaN material system.

1.2 Research problems

To extend VCSEL emission to the ultraviolet region, the challenges lie in many aspects.

The first important issue is the active region, which is the basis no matter for the LEDs, laser

diodes or VCSELs. For the DUV wavelengths, AlGaN which has high energy bandgap

has to be used. But there is a large lattice-mismatch between the layers and commercial

substrates such as sapphire and SiC. The structures grown on foreign substrates always

have high dislocation density that leads to low internal quantum efficiency (IQE). Bulk AlN

substrates with low dislocation density (below 108 cm−2) can be used, but they suffer from

high impurity absorption, high cost and limited availability. Epitaxial lateral overgrowth

(ELO) on sapphire can be used to reduce dislocations [18–21], but it requires complex

processing steps such as cleanroom fabrication (lithography and etching) and regrowth of

thick layers, which would increase material and time costs.

Besides the quality of substrates, there is also a degradation of the structural quality of

epitaxial AlGaN layers with increasing Al molar ratio. The impurity effect is greater in

AlN than GaN since Al atoms have small diffusion length (higher sticking coefficient) and

increased affinity to oxidize than Ga atoms, which cause rather high density of defects in

AlGaN layers with Al content higher than 50% [22, 23].

Thirdly, c-plane nitrides exhibit piezoelectric and spontaneous polarization, which might

be useful for piezo-devices but leads to strong quantum-confined Stark effect (QCSE) sep-

arating electrons and holes in the MQWs. This phenomenon can significantly reduce the

emission efficiency. C-plane AlGaN alloys also exhibit optical polarization anisotropy: E-

field ⊥ c polarized emission (which will be referred to as TE polarization) decreases when

compared to the emission polarized along the c axis (E-field ‖ c) as the Al composition

increases for the DUV wavelengths, which is detrimental to surface emission.

12

For current-injected DUV devices, effective doping is difficult to achieve, especially for

Al(Ga)N with wide energy bandgap. Si is commonly used as n-dopant. However, Si-doped

AlxGa1−xN layers typically show dramatically increasing resistivity for x>0.70 [24–27] and

it’s also crucial to obtain effective p-type conductive AlGaN with high Al content. These

are due to high donor/acceptor activation energy and compensation effects [27, 28]. So in

this work, optical pumping is used to avoid electrical issues.

For the VCSEL structure, another big challenge is the DBRs. In the DUV range, the

conventionally used AlGaN/AlN DBRs have smaller refractive index contrast leading to

low reflectivity and narrow stopband. The large lattice mismatch between the epitaxial

structures and the substrates would give rise to strain-related defects. So, high-reflectivity

DBR structures with large stopband width for the UV and DUV spectra need to be further

developed.

In the framework of developing DUV VCSELs, this work has be divided into several

tasks regarding the different challenges mentioned above that include the realization of

AlGaN MQW structures for 280 nm with enhanced oscillator strength of surface emission,

MOVPE growth study of new BAlN material, as well as simulation and realization of DUV

BAlN/Al(Ga)N DBRs. These tasks are for the processing of final light-emitting devices

such as resonant-cavity LEDs (RC-LEDs) and VCSELs.

1.3 State-of-the-art

Visible and UV lasers based upon the wide bandgap semiconductors including ZnSe- and

GaN-based material systems have been extensively explored for the potential applications

of high density optical storage systems, laser printer engines, full color display systems,

and large-area projector systems. The ZnSe-based materials were the first material sys-

tem to provide continuous wave operation of blue-green edge-emitting lasers (EELs) [29].

However, serious reliability problems possibly arising from its chemically and structurally

13

unstable material system have hindered the application of ZnSe lasers in the real-world sys-

tems. Later, AlGaInN-based wide bandgap semiconductor materials [30, 31] have brought

innovative changes in photonic devices, since their strong chemical bonds bring in out-

standing chemical and thermal stability, and their direct bandgaps can cover from visible

to DUV spectra [32], as shown in Fig. 4. Especially in 2014, the Nobel Prize in physics

was awarded to Professor Akasaki, Professor Amano, and Professor Nakamura for their

important contributions to bright blue LEDs in the early 1990s [33].

For UV devices, the first nitride-based injection laser centered at around 390 nm was

demonstrated in 1995 by the group of Professor Akasaki [34]. AlGaN-based UV LEDs

for wavelengths shorter than 360 nm were initiated in 1998 [35]; then UV and DUV light

devices based on AlGaInN have undergone tremendous evolution through rapid progress

in material growth, device fabrication and packaging. In the DUV range, the LEDs and

EELs have been widely reported. The group of Sensor Electric Technology has reported

high external quantum efficiency (EQE) of 10.4% at 20 mA continuous current with output

power up to 9.3 mW for encapsulated AlGaN LEDs emitting at 278 nm in 2012 [36]. UV

Craftory has developed commercial production of 50 mW AlGaN LEDs with wavelength

ranging from 255 nm to 355 nm which have more than 10% EQE and over 10000 hour

life time [37]. The reported EQE of UV LEDs as a function of the emission wavelength

is shown in Fig. 5. Compared with blue LEDs, the EQE decreases rapidly when going

towards shorter wavelengths due to high defect density, low extraction efficiency and other

problems mentioned in the section 1.2.

For laser diodes or EELs, electrically-injected devices were mostly operating at 320∼400

nm. The Palo Alto Research Center obtained lasering at 368 nm with a threshold of 13

kA/cm2 and maximum output of 300 mW (6.7% for efficiency) [38]. Meijo University

demonstrated a threshold of 12.2 kA/cm2 at 356 nm [39]. The group CRHEA-CNRS re-

ported laser diode at 394 nm with a threshold of 6.7 kA/cm2 [40]. Hamamatsu Photonics

reported laser diodes at 342 nm and 336 nm with threshold of 9 and 18 kA/cm2 [41]. The

14

Figure 4: Bandgap energy versus in-plane lattice parameter diagram for III-nitrides.

progress of shorter wavelength devices has been limited by the large dislocation densities

and residual strain leading to quality degradation and high optical cavity loss [42].

In DUV region, most of the laser diodes reported are under optical pumping. The group

from Technische Universitat Berlin has reported the laser grown on the AlN bulk substrate

emitting at 279 nm with a threshold of 50 mJ/cm2 (IQE 20∼30%), and the laser grown

on ELO (epitaxially laterally overgrown) AlN/sapphire templates emitting at 272 nm with

a threshold of 65 mJ/cm2 (IQE 10∼20%). Both are TE-polarization dominant [21]. The

Hexatech group reported low threshold power of 84 kW/cm2 for lasing at 280.8 nm [43] and

Palo Alto Research Center demonstrated lasing at 266 nm with a threshold of 41 kW/cm2

[44]. Both devices are based on the AlGaN MQWs grown on single crystal AlN bulk

substrates. To achieve internal quantum efficieny (IQE) higher than 60%, AlN bulk single

crystal can be used since its threading dislocation density is below 5 × 108 cm−2 [45], but

it suffers from high impurity absorption, high cost and limited availability. RIKEN group

has used ammonia pulsed-flow multilayer growth to fabricate AlN templates on sapphire

to obtain IQE of 60% from AlGaN QWs [45, 46]. Recently, for the devices grown on AlN

templates on sapphire, simulated emission was observed at wavelengths of 256 nm and

15

Figure 5: External quantum efficiency as function of wavelength reported at the Interna-tional Workshop on Nitride Semiconductors in 2012 and ICNS 2013 [2].

249 nm with thresholds of 61 kW/cm2 and 95 kW/cm2 at room temperature, as reported by

the group of Prof. Dupuis from Georgia Institute of Technology. [47]. It is also reported

by Institute of Semiconductors (Chinese Academy of Sciences) that silicon doping could

effectively reduce the lasing threshold of UV lasers, and the stimulated emission at 288

nm for the structure grown on AlN/sapphire was obtained at the optical-pumping threshold

energy density of 64 mJ/cm2 (IQE = 42%) [48]. The state-of-the-art optically-pumped

AlGaN DUV lasers have been summarized in Fig. 6.

It is worth noting that very recently a breakthrough has been achieved by McGill uni-

versity. They reported the first 210 nm emitting AlN nanowire LEDs on Si substrates with

a turn on voltage of about 6 V, which is significantly lower than the commonly observed

20∼40 V due to efficient Mg doping [49]. They reported the first electrically-injected Al-

GaN lasers on Si operating at 320∼340 nm with ultra-low threshold (tens of A/cm2) [42] by

using AlGaN core-shell nanowires. Then, the electrically-injected AlGaN nanowire lasers

grown directly on Si substrates were achieved, which operates at 262.1 nm at 77 K with a

16

Figure 6: State-of-the-art optically-pumped AlGaN DUV lasers.

threshold of 200 A/cm2 [50]. The nanowire structure provides an avenue for electrically-

injected DUV lasers.

The development of III-nitride-based VCSELs includes optically-pumped devices at

blue wavelengths [51, 52] and room temperature (RT) current-injected VCSELs at blue/violet

wavelengths [53–55]. Redwing et al. from Advanced Technology Materials [51] devel-

oped an optically-pumped VCSEL based on GaN active region sandwiched between 30-

period Al0.4Ga0.6N/Al0.12Ga0.88N DBRs. A narrow laser mode appeared at 363 nm and

the threshold was around 2 MW/cm2 at room temperature. Krestnikov et al. from A.F.

Ioffe Physico-Technical Institute (Russian Academy of Science) reported photopumped In-

GaN/GaN/AlGaN VCSEL at the wavelength of 401 nm and 415 nm with thresholds of 400

kw/cm2 and 550 kw/cm2. A 2λ vertical cavity included twelvefold stacked InGaN inser-

tions in a GaN matrix grown on top of 37-period Al0.15Ga0.85N/GaN DBR [52] and GaN/air

interface acted as top mirror. Higuchi et al. from Nitride Semiconductor Research Labora-

tory in Japan demonstrated continuous wave (CW) lasing VCSEL at 414 nm with threshold

current of 7.0 mA at RT [53]. It was mounted on a Si substrate by a wafer bonding and the

17

sapphire substrate was removed by laser lift-off. The active region consisted of a 2-pair In-

GaN/GaN QWs. 7λ GaN cavity was embedded between two SiO2/Nb2O5 DBRs. Recently

in 2014, Lin et al. from National Chiao Tung University in Taiwan showed an InGaN VC-

SEL, in which the application of composition-graded electron blocking layer reduced the

threshold from 12.6 to 9.2 kA/cm2 and enhanced output power by a factor of 3.8 [54]. So

far, no efficient VCSELs operating below 300 nm have been reported.

One of the biggest challenges for UV VCSELs lies in the need for high-reflectivity

DBRs. Conventional AlGaInN-based DBRs have been explored to give high reflectivity

from blue to the near-UV region. For example, 25-pair Al0.18Ga0.82N/Al0.8Ga0.2N with a

reflectivity as high as 99% and a stopband width of 26 nm at central wavelength of 347 nm

was reported by Mitrofanov et al. from Bell Laboratories [56]. Feltin et al. from Ecole

Polytechnique Federale de Lausanne also used Al0.2Ga0.8N/Al0.85In0.15N lattice matched

system to achieve 99% reflectivity at 340 nm [57]. For the DUV wavelengths, the DBRs

based on conventional AlGaInN material system are shown in Tab. 1. Moe et al. from

University of California - Santa Barbara demonstrated reflectivity of 66.6% and 82.8% for

the wavelengths of 245 nm and 279 nm, using 10-period Al0.63Ga0.37N/AlN and 21-period

Al0.58Ga0.42N/AlN, respectively [58]. Getty et al. from the same group achieved 80% re-

flection at 254 nm and 280 nm by using 19-period Al0.60Ga0.40N/ Al0.95Ga0.05N [59]. Zhang

et al. from Nanjing University in China obtained 83.9% reflection at 246 nm by using

13-pair Al0.77Ga0.23N/Al0.98In0.02N [60], and 68.8% reflection at 247 nm by using 11-pair

Al0.70Ga0.40N/AlN [61]. Very recently, Hoffmann et al. from North Carolina State Univer-

sity reported the highest reflectivity of 97 % at 270 nm by using 24.5-pair AlN/Al0.65Ga0.35N

strain-compensated DBRs on Al0.85Ga0.15N relaxed buffer [62].

It is difficult to achieve high reflectivity requirement due to large absorption and small

refractive index contrast of the materials in DUV region. Conventionally-used AlGaN lay-

ers exhibit lattice mismatch as high as ∼2.4% between GaN and AlN to achieve only a

small refractive index contrast in the mirror structure. The high reflectivity requiring large

18

Table 1: State-of-the-art distributed Bragg mirrors based on AlGaInN for wavelengths be-low 300 nm.

References Structures Reflectivity Stopband width[58] 10-pair Al0.63Ga0.37N / AlN 66.6% R at 245 nm ∼20 nm[58] 21-pair Al0.58Ga0.42N / AlN 82.8% R at 279 nm ∼18 nm[59] 19-pair Al0.60Ga0.40N / Al0.95Ga0.05N 80% R at 254 nm ∼10 nm[59] 19-pair Al0.60Ga0.40N / Al0.95Ga0.05N 80% R at 280 nm ∼17 nm[60] 13-pair Al0.77Ga0.23N / Al0.98In0.02N 83.9% R at 246 nm ∼18 nm[61] 11-pair Al0.70Ga0.40N / AlN 68.8% R at 247 nm ∼18 nm[62] 24-pair Al0.65Ga0.35N / AlN 97% R at 270 nm ∼10 nm

number of pairs was accompanied by the dislocations, cracks or rough interfaces. So the

novel system needs to be considered to build efficient DBR structures below 300 nm. In

this thesis, BAlGaN materials will be studied. The primary reason for the B incorporation

is that a very small amount of B in AlN could introduce a strong refractive index con-

trast [63, 64]. For example, at the wavelength of 280 nm, the Al0.65Ga0.35N/AlN system

has refractive index contrast of 0.07 with lattice mismatch as large as 0.8%. Meanwhile,

BAlN with only 1.2% B can have a refractive index contrast of 0.17 with AlN and lattice

mismatch is as small as 0.2%. Besides, BAlN system exhibits less optical absorption than

AlGaN due to its large bandgap. Additionally, strain-compensated structure could be de-

signed by alternating BAlN and AlGaN layers in the DBR so that both the large refractive

index contrast and lattice matched structure could be obtained at the same time. By us-

ing this material, our group has obtained 60% reflection at 282 nm by 18-pair BAlN/AlN,

82% reflection at 311 nm by 24-pair BAlN/AlN [64] and 92% at 360 nm by replacing

AlN with AlGaN. DBRs based on different nitride systems including Al(Ga)N/Al(Ga)N,

AlInN/Al(Ga)N, BGaN/GaN and BAlN/Al(Ga)N are summarized in Fig. 7.

The study of boron containing materials can be very helpful for III-nitride devices since

boron can bring additional freedom in engineering the bandgap, lattice constant and refrac-

tive index of multi-layered structures. For example, besides the enhanced refractive index

contrast mentioned earlier, the BAlGaN system can decrease or eliminate lattice mismatch

on SiC and AlN substrates [65–67]. However, further investigation of the basic epitaxial

19

Figure 7: DBR reflectivity for various nitride systems.

growth is required for this new material. BAlN and BGaN have been grown by MOVPE

[63, 64, 68–75], MBE [67, 76, 77] and magnetron sputtering [78, 79], but the boron con-

tent and crystallinity are limited due to phase separation, short diffusion length of boron and

strong parasitic reaction in the gas phase [68, 72, 76, 78, 80]. By magnetron sputtering, Lil-

jeholm et al. from Dev. Solid State Electronics in Sweden achieved high crystalline quality

of BAlN with 18% boron [78]. By MOVPE method BAlN materials have normally been

grown above 1050 ◦C [63, 68, 69, 75], or at 1000 ◦C by using flow-modulate epitaxy (FME)

[72, 81] in order to enhance migration of boron and aluminum atoms. But clear XRD peaks

of wurtzite-BAlN layers reported are related to only 1∼2% boron [63, 68, 72, 75]. Shibata

et al. from Kohgakuin University in Japan [69] showed high boron content incorporation

by MOVPE, but the related XRD peak intensity was very weak. Further study for BAlN

with high boron content has to be established for its big potential in applications such as

MQWs, DBRs, and sensors.

20

1.4 Scope of the thesis

The objective of the project is for making progress on VCSELs operating below 300 nm

based on the BAlGaN materials. 280 nm is targeted since it is a common wavelength used

in the market for DUV applications such as sterilization or water purification. The VCSEL

structure consists of dielectric top mirror, AlGaN active region and bottom DBR, and this

work contributes to the two main parts: AlGaN-based active region and BAlGaN-based

DBRs.

Chapter 1 introduces the background of the project. The motivation of the project in-

cluding possible applications is stated. The research problems in different aspects regard-

ing this topic are described. The state-of-the-art of different types of light emission devices,

DBRs for DUV wavelengths and progress of BAlN is also summarized in this chapter.

Chapter 2 presents the basic properties of III-nitride material, and a brief introduction

of the techniques and equipments used for the growth and characterizations is included.

Chapter 3 is devoted to the MOVPE growth studies of AlGaN and BAlN on AlN/sapphire

templates. Firstly, AlGaN single layers with different composition and relaxation are cal-

ibrated. Based on these samples, the relationship between composition and strain state is

discussed. Different models are brought in to predict the range of critical thickness of Al-

GaN grown on AlN. The method to calculate screw/edge threading dislocation density of

AlGaN and AlN layers based on XRD measurements is also established. Secondly, for the

studies of novel BAlN material, the characteristics of BAlN with high boron content (more

than 10%) are analyzed. In order to improve the crystalline quality of BAlN, a method of

low-temperature growth with annealing is proposed. The influences of growth temperature

and TEB/III ratio on the boron composition and on the crystalline quality are investigated.

Chapter 4 is addressed to the AlGaN MQWs emitting at 280 nm. The design of MQWs

is to enhance TE-polarized emission. The structural and optical properties of 4-, 10- and

20-period MQWs are studied here. Besides, the analysis of typical defects in QW samples

is also presented in this chapter.

21

Chapter 5 focuses on the theoretical simulation and the experimental realization of

BAlN/Al(Ga)N DBRs. For the reflector software developed before in our group, differ-

ent factors leading to degradation of DBR performance such as surface roughness, interface

roughness and strain are introduced into the simulations, and their influences are discussed.

Then a series of BAlN/Al(Ga)N DBRs are grown by MOVPE and analyzed by different

characterization methods.

Chapter 6 is divided into three sections. Firstly, the conclusion part summarizes the re-

sults of this thesis. Secondly, the further possible research directions for optimizing MQWs

and DBRs as well as achieving final devices are proposed in the perspective part. The pub-

lications and awards achieved during the time of Ph.D study are listed in the end of the

chapter.

22

CHAPTER 2

FUNDAMENTALS OF MATERIALS AND EXPERIMENTS

2.1 Fundamental properties of III nitrides

The III nitrides are excellent wide bandgap semiconductors with outstanding features. They

have enabled a revolution in modern electronic and optoelectronic applications. The basic

properties of III-nitrides are described in the following sections, which are the fundamentals

in the aspect of materials used in this work.

2.1.1 Structural properties of III nitrides

III-nitrides including binary, ternaries and quaternaries of (B, Al, Ga, In)N have two types

of crystalline structures: hexagonal and cubic. Under the typical MOVPE growth condi-

tions, they crystallize in the most thermodynamically stable form into wurtzite structure,

except that the BN favors hexagonal (graphitic) prototype. As shown in Fig. 8, wurtzite

crystallographic structure has nitrogen atoms forming a hexagonal close packed structure

while the group III elements occupying half of the tetrahedral sites available in the lat-

tices. Or it can be seen as two interpenetrating hexagonal close packed sub-lattices, and

each sub-lattice is shifted along c-axis by 3/8 of the cell height. Each atom in the lattice

is tetrahedrally coordinated. Typical crystal orientations and planes of wurtzite III-nitrides

are illustrated in Fig. 9.

Since there is a large disparity of electronegativity between III elements and nitrogen

atoms with small atomic radius, the III elements and nitrogen are bound by strong covalent

bonds. Therefore, III-nitrides are chemically and physically stable. III-nitrides also show

high thermal conductivity as seen in Tab. 2. Owing to their structure and thermal stability,

III-nitrides are suitable candidates for high temperature and high power applications. The

lattice parameters of hexagonal BN, AlN, GaN and InN are summarized in Tab. 2. The

23

Figure 8: Wurtzite crystal structure of III nitrides [3].

Figure 9: Typical crystal orientations and planes of wurtzite III-nitrides.

24

Table 2: Properties of III-nitrides binaries [1].Parameters BN AlN GaN InN

a(Å) 2.55 3.122 3.189 3.548c(Å) 4.17 4.982 5.185 5.718

Eg (eV) 5.9 6.28 3.42 0.7Thermal expansion (4a/a) (10−6k−1) -2.7 4.15 5.59 3.8Thermal expansion (4c/c) (10−6k−1) 38 5.27 3.17 2.9Thermal conductivity (W·cm−1k−1) 6 (‖a), 0.3 (⊥a) 2.85 1.3 0.45

lattices of ternaries can be calculated by Vegard’s law [82]:

a, c(AxB1−xN) = a, c(AN) · x + a, c(BN) · (1 − x) . (1)

One of the technical challenges of III-nitride applications is the lack of lattice-matched

substrates. Commercially available, low-cost and thermal stable sapphire substrates were

used in the majority of the research work. Although it has large lattice mismatch with III-

nitrides, GaN templates grown on sapphire have already been achieved with the dislocation

density down to ∼108 cm−2 by using low temperature nucleation layer, which was pro-

posed by Nakamura in 1991 [83], and they have been widely used for visible and near-UV

light emitting devices. For DUV devices, the high quality AlN substrates are still under

investigations now and the method to reduce dislocation densities is an important topic.

2.1.2 Optical properties of III nitrides2.1.2.1 III nitride bandgaps and band crossover of AlGaN

III-nitrides have wide bandgaps spanning a wide range from 0.7 to 6.2 eV (as shown in

Fig. 4 and Tab. 2), especially the BAlGaN covers the UV and DUV region. The energy

bandgaps of ternaries can be calculated directly from their binaries:

Eg(AxB1−xN) = Eg(AN) · x + Eg(BN) · (1 − x) − b · x · (1 − x) . (2)

In this equation, “b” represents the energy bandgap bowing parameter. Tab. 3 sum-

marizes some bowing values of III-nitrdes. In fact, there is a large spread and uncertainty

of values for bowing parameters in the literature. For example, the bowing parameter of

25

Table 3: Bandgap bowing parameters of III-nitrides ternaries.BAlN BGaN AlGaN AlInN InGaN

Bowing (eV) 5.45 [86] 9.2 [74] 1 [87] 2.5 [87] 3 [87]

1 eV is often used in band calculations of AlGaN. Meanwhile, Lee et al. recommended a

value of 0.6 eV for intrinsic bandgap bowing of AlGaN. They found that the AlGaN lay-

ers directly nucleated or buffered on sapphire at high temperatures usually lead to a strong

downward bowing of at least 1.3 eV, and it also tends to jump to strong bowing as Al frac-

tion increases [84]. The apparent strong bowing might be an artifact resulting from defect-

or impury- related transitions at energies below the bandgap [85].

Few results have been reported on bowing parameters of boron-based III-nitrides. For

BGaN material, the value of 9.2 eV has been experimentally obtained by group of Ougaz-

zaden [74]. For BAlN, there is only theoretical calculations showing that BAlN has strong

bowing parameter of 5.45 eV [86]. Since BAlN material is not as well studied as other

III-nitrides, it still requires more experimental investigations for its optical properties.

In AlGaN alloys which constitute the active region of DUV light emitting devices, the

order of the two upper valence bands is modified with respect to binary GaN. Apart from

the modification of the electric band structure, alloying also leads to a modification of

the optical polarization properties of the interband transitions, which influences the perfor-

mance of the devices. In GaN, the crystal field term ∆ is positive, so the upper valence band

corresponds to the heavy hold band of Γ9 symmetry, while the second one corresponds to

the split-off hole band of symmetry Γ7. However, the ∆ term for AlN is negative with Γ7

symmetry becoming the upper band. In this case, along with the increase of Al content

in AlGaN, the energy order of heavy hold band and split-off hole band reverses at a cer-

tain composition (Fig. 10). The crossover point would be influenced by the parameters

used in the theoretical calculations as well as the strain state of the layer [88, 89]. For

example, Nam et al. reported a crossover point of xAl=0.25 [88] while Leroux et al. ob-

tained a crossover composition of xAl=0.1 [89]. This energy band crossover between AlN

26

Figure 10: Band alignment of GaN/AlGaN structure [4].

and GaN is also accompanied by a switching of the valence band state symmetry. The

topmost Γ7 valence band is governed by pz-like state. The following bands (Γ9 and lower

Γ7) are governed by px and py-like states. The oscillator strength of the optical transitions

is then modified and consequently the interband absorption coefficients (and stimulated-

emission coefficients) are changed. In particular, the oscillator strength between the con-

duction band and the upper Γ7 valence band, which becomes the fundamental transition, is

highly reduced with increasing Al fraction for the configuration where the electric field is

perpendicular to the c axis [88–92]. This effect is detrimental to the surface emission of

AlGaN-based DUV LEDs and lasers, and it needs to be considered during the design of the

devices. The band structure and alignment are detailed in the section 4.1.1.

2.1.2.2 Polarization and quantum-confined Stark effect

III-nitrides display strong spontaneous and piezoelectric polarization due to their wurtzite

crystal structure and high degree of ionicity. The polarization-induced electric fields can

cause strong quantum-confined Stark effect (QCSE) in the QWs and reduce significantly

27

Table 4: Polarization elastic parameters from the literature.AlN GaN InN

Psp [C/m2] -0.081 [93] -0.029 [93] -0.032 [93]e33 [C/m2] 1.46 [93] 0.73 [93] 0.97 [93]e31 [C/m2] -0.60 [93] -0.49 [93] -0.57 [93]

ε 10.7 [97] 10.4 [98] 14.6 [99]C13 [GPa] 94 [100] 68 [100] 70 [100]C33 [GPa] 377 [100] 354 [100] 205 [100]

the internal quantum efficiency.

The total polarization is described as the sum of spontaneous polarization and piezo-

electric polarization:

P = Psp + Ppz . (3)

Spontaneous and piezoelectric polarization of ternary alloys could be interpolated by two

end points (Vegard’s law). Taking AlxGa1−xN as an example, two types of polarization can

be expressed as:

Psp(AlxGa1−xN) = xPsp(AlN) + (1 − x)Psp(GaN)

Ppz(AlxGa1−xN) = xPpz(AlN) + (1 − x)Ppz(GaN) .(4)

Psp for each binary could be obtained from literatures, and Ppz is related to the strain of the

layer described as following equations [5, 93–96]:

Ppz = ei jε j

P3 = e33 + e31(ε1 + ε2) = 2a − a0

a0(e31 − e33 ·

C13

C33) ,

(5)

where ei j is piezoelectric coefficients, ε j is strain tensor, e and C are the piezoelectric coef-

ficients and elastic constants, respectively. Parameters a and a0 are in-plane lattice param-

eters under the strain and free of strain. The polarization and elastic parameters from the

literature are summarized in Tab. 4.

The orientation of spontaneous/piezoelectric polarization is indicated in Fig. 11. The

spontaneous polarization field for GaN, AlN and InN (0 0 0 1) films are in the [0 0 0 -1]

28

Figure 11: Spontaneous polarization field (Psp) and piezoelectric polarization field (Ppz) forGaN, AlGaN and InGaN coherently strained to GaN (0 0 0 1) [5].

direction. The relevant piezoelectric field of (0 0 0 1) films is in the [0 0 0 1] direction

under compressive strain and in the [0 0 0 -1] direction under tensile strain [5].

For MQWs, the conservation of the electric displacement vector across the heterostruc-

ture leads to:

εwEw − εbEb = Pb − Pw , (6)

where εw (εb) is the dielectric constant of well (barrier), Ew (Eb) is the electric field in

the well (barrier), and Pw (Pb) is the zero-field polarization of the well (barrier). After

combining the boundary condition:

LwEw + LbEb = 0 , (7)

where Lw and Lb are the widths of the well and the barrier, the built-in electric field could

be deduced from polarizations [95, 101]:

Ew =Lb(Pb − Pw)

(εbLw + εwLb). (8)

The electric field in the QWs leads to the bending of energy bandgap: the electron states

shift to lower energies, while the hole states shift to higher energies. The effective energy

bandgap is reduced resulting in longer wavelength. Additionally, the electric field shifts

electrons and holes to the opposite side of the well, decreasing the overlap of wavefunc-

tions [102]. This effect is called quantum-confined Stark effect (QCSE). In the simplest

approximation, the photoluminescence energy EPL can be expressed as:

EPL = Ee + Eh + Eg − Eexc − eFL , (9)

29

where Eg is the energy bandgap, Eexc is the exciton binding energy, Ee and Eh are energies

of the first quantum levels in the well for electrons and holes, F is the total electric field

and L is the thickness of the quantum well. In AlGaN/AlGaN MQWs, the emission wave-

length and recombination efficiency are strongly affected by built-in electric field E and the

thickness of the wells. By reducing the thickness of the wells, QCSE can be alleviated.

2.2 Experimental methods2.2.1 Metal-organic vapor-phase epitaxy

Metal-organic vapor-phase epitaxy (MOVPE) is one of the methods utilized to epitaxi-

ally grow semiconductor materials. Compared with other techniques such as hydride va-

por phase epitaxy (HVPE) or molecular beam epitaxy (MBE), the strengths of MOVPE

technique include high growth rate, versatility, high quality, and suitability for large-scale

production, which make it become a major process in the manufacture of optoelectronics.

2.2.1.1 Basic principles

MOVPE growth is governed by the diffusion processes, and it is conducted under near

thermodynamic equilibrium conditions that rely on the vapor transport of precursors in a

heated zone [103]. The sources of group III elements are either liquids (trimethylgallium,

trimethylaluminum, thriethylgallium, triethylborane) or solids (trimethylindium). They are

stored in bubblers that are maintained at a constant temperature. The carrier gas (nitrogen

or hydrogen) flows into the bubblers, saturates with vapor of the sources and transports

them into the reactor. The schematic of III-nitride epitaxial growth is illustrated in Fig.

12. The wafer is heated in the reactor chamber to create near equilibrium conditions. The

organometallic precursors are mixed with ammonia near the inlet of the reactor and are

then transported to the heated zone. The heated organic precursor molecules decompose in

the absence of oxygen or any halogen (pyrolysis). The resulting species diffuse through the

boundary layer to the growing surface and attach onto the substrate by physisorption. The

species can desorb or react with other surface species at this moment. To form a new layer,

30

Figure 12: Schematic of III-nitrides epitaxial growth.

the species can diffuse on the surface and bound tightly at the bottom of a growth step, or

nucleate at other positions to form islands. The general reaction is described in the Eq. 10:

R3M + NH3 � MN + 3RH . (10)

The byproducts that form during the deposition reactions diffuse into the carrier gas

away from the deposition zone and flow into the reactor exhaust. Parasitic reactions be-

tween precursors can also occur in the gas phase which would reduce the incorporation

efficiency of sources and degrade the quality of epitaxial layers since the particles of the

byproducts may fall on the substrate surface hindering the formation of single crystal.

MOVPE growth is a complex process including thermodynamics, kinetics, hydrody-

namics and mass transport. The theoretical details can be found in [103]. Generally speak-

ing, there are several basic growth parameters which can be adjusted to control the growth

rate and the quality of epitaxial layers: temperature, pressure, V/III ratio (the ratio of am-

monia flow rate to the total flow rate of organometallic precursors), III/III ratio (the concen-

tration of one type organometallic precursor in the total organometallic precursors), carrier

gas.

Growth temperature should be optimized for the epitaxy of different alloys. On one

31

hand, it requires energy for the pyrolysis of precursors, the diffusion of the atoms on the

substrate and activation of the reactions, which is important for the growth rate and surface

structural property. On the other hand, at very high temperature, the desorption would be

dominant and the substrate surface can be decomposed. Additionally, high temperature

would lead to strong parasitic reactions between precursors. Furthermore, growth temper-

ature can strongly shift the composition of In in III nitrides such as InGaN and AlInN.

MOVPE growth of III nitrides is normally done under low pressure (13-400 mbar), in

order to reduce the parasitic reactions.

V/III ratio is important for the quality of the layers. If the V/III ratio is too low, supply

of nitrogen atoms would be inefficient, and nitrogen vacancies and auto background doping

would be increased. But if the V/III ratio is too high, parasitic reactions would be aggra-

vated and surface mobility of adsorbed atoms would be impeded. III/III ratio which means

the concentration of one precursor in the total precursors for III elements (such as TMAl/III

and TMG/III) is the key parameter to adjust the composition of the ternary or quaternary

alloys. But under the same gas-phase concentration, the composition can also be shifted

due to other parameters including temperature, pressure, or the change of the strain state

(composition pulling effect).

The carrier gas used in this work is H2, since it has beneficial thermal conductivity,

faster diffusivity of precursor species and carbon-radical scouring properties compared with

N2. It should be noted that the presence of hydrogen atoms is detrimental to InGaN growth

since it enhances the In desorption and it is harmful to Mg-doped layers due to the forma-

tion of Mg-H complexes.

2.2.1.2 Equipment and in-situ characterizations

Epitaxial materials and structures studied within this work were grown by low-pressure

metal-organic vapor-phase epitaxy (LP-MOVPE) using a home-made system if not spec-

ified. This system was designed and installed by Prof. Abdallah Ougazzaden [104], as

shown in Fig. 13. The system includes four basic elements: gas handling system, reactor

32

Figure 13: MOVPE system and T-shape reactor chamber.

Figure 14: Aixtron 3×2 inch, close coupled showerhead (CCS) MOVPE system.

chamber, heating system and an exhaust (low pressure) pumping system. The growth was

performed in a T-shape reactor [105]. The temperature range is from 400 ◦C to 1040 ◦C and

the pressure in the reactor can be regulated from 107 to 600 mbar for the growth. Hydrogen

or nitrogen can be used as carrier gas. The cold-wall system is applied so that the substrate

is much hotter than the other zone and the reactants can be depleted here contributing to the

growth. During growth, the substrate is rotated at 60 rpm to enhance the layer homogeneity

and to help maintain the laminar flow on the sample surface.

A new Aixtron 3×2 inch, close coupled showerhead (CCS) MOVPE system has also

been brought into operation and it is shown in Fig. 14. The surface temperature of sub-

strates in this reactor can reach 1300 ◦C. The future plans of optimizing AlN/sapphire

33

templates and improving the quality of BAlN layers will be performed in this new reactor.

The T-shape system is equipped with an in-situ optical reflectance monitor in order to

estimate the growth rate and get fist information of the surface quality of the sample.

Laser beam with wavelength of 633 nm is launched at the wafer and detected at normal

incidence. Interference between the reflection at the surface of thin films and reflection at

the bottom causes oscillations of detected light intensity as the layer is grown. The growth

rate can be calculated by the following equation:

Growth rate = λ/(2n4t) , (11)

where λ is the wavelength of the laser beam, n is refractive index of the layer and 4t is the

time interval between two maxima or two minima of oscillations.

During the growth of single layer, there would be additional light scattering due to the

surface roughness. The reduced reflection intensity of maxima or minima indicates the

increase of surface roughness, which can be expressed quantitatively as follows:

R = R0 · e−(4πσ/λ) , (12)

where R is the mean value of reflectance oscillations, λ is the used wavelength and σ refers

to the root mean square of roughness.

2.2.1.3 Precursors and calculations of debit

Trimethylaluminum (TMAl), thrimethylgallium (TMGa), triethylborane (TEB) and NH3

are used as precursors for aluminum, gallium, boron and nitrogen, respectively. TMG,

TMAl and TEB are stored in liquid and TMIn is in solid. The containers, or referred as

“bubblers”, are maintained in the bath with a set temperature. A certain amount of carrier

gas is injected into the bubbler. The vapor phase of the source saturates in the carrier gas,

and then it is carried out of the bubbler and delivered into the reactor. So the debit of the

sources can be calculated by the following equation:

Debit(µmol/min) =N(sccm)

22400(ml/mol)·

Psat(Torr)Pbubbler(Torr)

· 106, (13)

34

Table 5: Properties of metal-organic precursors of III elements.TEB TMAl TMG TMIn

A/B 7.413-1544.2

T − 27.428.22/2134 8.07/1703 10.52/3014

Melting temperature (◦C) -93 15 -15 88Boiling temperature (◦C) 95 125 56 134

where N is the flow rate of the carrier gas into the bubblers which is measured and con-

trolled by a mass flow controller in the inlet of each bubbler. Pbubbler is the pressure inside

the bubbler which is measured and regulated by a pressure regulator. Psat is the equilibrium

vapor pressure of the organometallic source, which can be tuned by the bubbler temperature

T (K) and has a general form as:

Psat(Torr) = 10A−B/T . (14)

The parameters used in the calculations and physical properties of precursors are summa-

rized in Tab. 5

The debit of NH3 flowing into the reactor can be directly calculated from the flow rate:

Debit(µmol/min) =NNH3(slm)

22.4(l/mol)· 106. (15)

Based on the debit of each precursor, the V/III ratio and III/III ratio can be easily obtained.

2.2.2 Characterization techniques

After growth, various kinds of characterization techniques are required for acquiring the

properties and the quality of grown materials and devices.

• X-ray Diffraction (XRD)

X-ray diffraction is an effective method to determine composition and provide structural

information of single epitaxial layers as well as complex multi-layered structures. The basic

principle is the Bragg’s law in Eq. 16:

2 · dhkl · sinθ = n · λ . (16)

35

The incident X-ray would be reflected on a set of crystal planes. If the incident angle θ and

the crystal plane distance dhkl satisfy this equation, there would be constructive interference

to obtain a diffraction peak. X-ray diffraction allows to analyze accurately the lattice pa-

rameters, hence the composition and strain state. The X-ray can also have reflection on top

and bottom of each layer, so that it can give an accurate measurement of thickness for thin

films or MQWs. Since the dislocations cause small disorientation (tilt or twist) of crystal

blocks leading to the broadness of rocking curves, the screw and edge dislocation densi-

ties can be estimated by this technique. More details about principles and equations for

calculations can be found in Ref. [106].

In this work, high resolution XRD measurements were performed in a Panalytical

X’pert Pro MRD system with Cu Kα radiation (Cu Kα1 : 1.5405 Å). The height of X-

ray beam from hybrid monochromator is 1.2 mm and the resolution is ∼ 12 arcsec.

• Secondary ion mass spectroscopy (SIMS)

Secondary ion mass spectroscopy is a technique used to analyze the composition of

solid surface or thin films by sputtering the surface of the specimen with focused primary

ion beam, collecting and analyzing ejected secondary ions. The mass/charge ratios of these

secondary ions are measured with a mass spectrometer to determine the elemental, isotopic,

or molecular compositions [107]. It is a sensitive technique to detect low concentration of

atoms such as impurities or dopants. The depth concentration profiles for different elements

can be obtained by plotting concentrations versus sputtering depth.

In this work, SIMS was used to determine the content of boron, which is a major ele-

ment in the alloy. Estimation of major elements composition requires the so-called MCs+

molecular ions (where M is the element to be monitored) technique to minimize the matrix

effect. It is necessary to dispose a standard made up of the same alloy whose composition

is already known. If the sample to analyze and the standard have very close compositions,

the quantification accuracy can be better than one percent [108]. To quantify boron content

36

in the BAlN layers, the simplified equation can be written as:

x/(1 + x) = K[B]/[Al] , (17)

where [B] and [Al] are the Cs+ signals of boron and aluminum, and K is the sensitivity co-

efficient deduced from the boron implanted standard. The error range by using this method

is estimated to be ±5%.

The SIMS measurements in this work have been performed by Probion Analysis located

in Bagneux (France).

• Scanning transmission electron microscope (STEM)

A scanning transmission electron microscope is a type of transmission electron micro-

scope (TEM). A beam of electrons is transmitted through an ultra-thin specimen. An image

is formed from the interaction of the electrons transmitted through the specimen. Owing

to the small de Broglie wavelength of electrons, this technique can examine the lattice of

atoms in the crystal. The difference of STEM from conventional TEM is that STEM fo-

cuses the electron beam into a narrow spot which is scanned over the the sample in a raster,

which makes it suitable for analysis such as energy dispersive X-ray (EDX) mapping, elec-

tron energy loss spectroscopy (EELS) and annular dark field imaging (ADF) allowing di-

rect correlation of image and quantitative data [109]. An annular dark field image, which is

formed only by very high-angle and incoherently-scattered electrons, is highly sensitive to

atomic number variations of the sample (Z-contrast images). By using this high-angle an-

nular dark-field scanning transmission microscopy (HAADF-STEM), the relative intensity

variations of the images reflect variations in the compositions of the material. Therefore,

HAADF-STEM images can be interpreted into quantitative compositional maps by using

EDX as a chemical calibration [110].

To prepare the samples for TEM characterizations, 100 nm carbon were deposited to

protect the surface. Then all the thin foils were prepared using focused ion beam (FIB)

thinning and ionmilling by Dr. David Troadec in Institut d’Electronique de Microelectron-

ique et de Nanotechnologie (IEMN, Lille, France). Carbon coating and HAADF-STEM

37

characterizations in this work were performed by Dr. Gilles Partriarche in Laboratoire de

Photonique et de Nanostructures (LPN, Marcoussis, France). The equipment for HAADF-

STEM is aberration-corrected JEOL 2200FS electron transmission microscope.

• Energy-dispersive X-ray spectroscopy (EDX)

The energy-dispersive X-ray spectroscopy is combined with TEM technique. The X-

rays produced when the electron beam impacts the structure are collected and measured

by an energy-dispersive spectrometer. As the energies of the X-rays are characteristic of

the energy difference between the two shells and of the atomic structure of the emitting

element, EDX allows the measurement of elemental compositions of the specimen [111].

• Atomic force microscope (AFM)

Atomic force microscopy is a non-destructive technique to measure the sample surface

in nanometer-scale resolution. It consists of a cantilever with a sharp tip typically made

of Si3N4 or Si. The tip radius of curvature is on the order of nanometers. When the tip is

brought into proximity of a sample surface, the interaction forces between the tip and the

sample cause a deflection of the cantilever according to Hook’s law. This deflection can be

measured by the reflection of the laser beam focused on the cantilever [112]. The motion

of the probe across the sample surface is controlled by feedback loop and a piezoelectronic

scanner moving the sample under the tip.

AFM has three primary modes: tapping mode, contact mode and non-contact mode.

The commonly used tapping mode maps topography by lightly tapping the surface with

an oscillating probe tip [113]. The oscillation frequency is equal or slightly lower than its

resonance frequency. In order to maintain a constant oscillating amplitude, the feedback

loop controls vertical position to maintain a constant tip-sample interaction. The vertical

position of the scanner is stored to form the topographic image of the sample surface.

In contact mode, the spring constant of the cantilever is lower than the effective spring

constant holding the atoms of most solid samples together. The contact force on the tip is

repulsive. The scanner gently traces the tip across the sample surface. By maintaining a

38

constant cantilever deflection, an image of the surface is obtained. In non-contact mode, a

stiff cantilever vibrates near the surface of the sample with the spacing on the order of tens

to hundreds ångstroms. The surface topography can be measured by monitoring changes in

the amplitude due to attractive van der Waal forces between the tip and the sample surface.

The AFM images in this work were obtained by Veeco 3100 Dimension Atomic Force

Microscope.

• Scanning Electron Microscope (SEM)

A scanning electron microscope is a type of electron microscope that produces images

of a sample by scanning it with a focused beam of high energy electrons [114]. The elec-

trons interact with atoms in the sample, producing signals of secondary electrons, back-

scattered electrons, characteristic X-rays, light (cathodoluminescence), Auger electrons,

transmitted electrons and phonons (heat). Secondary electrons are low energy electrons

emitted by atoms near the surface. The number of secondary electrons depend on the angle

at which the beam meets the surface of specimen. By scanning the sample and collecting

the secondary electrons with a special detector, the topography of the surface can be dis-

played. A wide range of magnifications from 10 times to more than 50 k times is possible.

High resolution (1 nm at 15 kV) makes it suitable for characterizing nano-structures.

The SEM images in this work were obtained by Zeiss supraT M 55VP. The main ele-

ments are: electron source, magnetic focusing lenses, the sample vacuum chamber, imag-

ing system and control panel.

• Cathodoluminescence (CL)

A cathodoluminescence spectroscopy combined with SEM system is used to study opti-

cal properties of the sample. The high energy electron bombardment onto a semiconductor

will result in the promotion of electrons from the valence band into the conduction band,

leaving behind a hole. When an electron and a hole recombine, it is possible for a photon of

characteristic wavelength to be emitted [115]. The CL emission is detected via a parabolic

mirror collector and analyzed by a spectrometer with a focal length of 320 mm using a 1200

39

grooves mm−1 grating with a spectral resolution of 0.06 nm. The signal is then registered

by a liquid N2-cooled HORIBA JOBIN YVON Instruments Symphony 1024 × 256 CCD

detector.

• Photoluminescence (PL)

The semiconductor sample is excited with a light source that provides photons with

an energy higher than the bandgap of the material. The photons will be absorbed and an

electron-hole pair will be created. The excited electrons undergo energy and momentum

relaxation towards the lowest energy state in the conduction band [116]. Then, the electrons

fall back to the valence band and recombine with the holes, emitting photons near the

bandgap energy.

The PL experiments are carried out under excitation provided by the third harmonic

generation of a Ti-Sapphire femto-seconde pulsed laser (245 nm) or by the second har-

monic beam of a continuous laser (266 nm). A continuous-flow cryostat can be used to

perform low temperature measurements. The emission was analyzed by a 1 m focal length

monochronometer and detected by a CCD camera. The laser beam is focused through a

microscope objective to a spot with diameter of approximately 2 µm. These measurements

were performed in Institut Pascal - Universite Blaise-Pascal (LASMEA, Aubiere, France).

• Fourier Transform Infrared Spectroscopy (FTIR) and optical spectrometer

FTIR spectrometer simultaneously collects data over a wide spectral range. An interfer-

ometer is applied to encode all the frequencies into a unique type of signal (interferogram)

[117]. To interpret and analyze the signal, the individual frequency can be decoded by

Fourier transformation. In this work, Fourier transform infrared spectroscopy of Bruker

Vertex 80V equipped with a deuterium lamp is used to measure the transmission (absorp-

tion) of the MQW structures in visible-DUV range. Absolute reflectance accessory uti-

lizing V-W geometry allows to measure the specular reflectivity of DBR structures. The

spectrum obtained by this mode is R2 (square of the reflectance) versus wavenumber.

40

The reflectivity and transmission spectrum of MQWs have also be measured by a spec-

trometer under excitation of a Xenon arc lamp in Institut Pascal UMR 6602 CNRS (LAS-

MEA, Aubere, France). The reflectivity and transmission signals are analyzed through the

same monochromator. The size of probed area is around 0.02 mm2.

41

CHAPTER 3

MOVPE STUDIES OF BAlGaN MATERIALS

This chapter focuses on the MOVPE growth and characterizations of AlGaN and BAlN

materials. The AlGaN material is a promising candidate for UV or DUV devices as well

as gas sensors. The novel BAlN exhibits great potentials for lattice, bandgap and refractive

index engineering. The studies of basic growth conditions and investigations of material

characteristics are required for further growth and fabrication of devices.

The active region of the laser or LED devices consists of AlGaN/AlGaN MQWs for

emission at DUV wavelengths. Therefore, a good control over composition and strain state

for AlGaN material is the foundation to realize the device design. So the first section of

this chapter focuses on the growth of AlGaN layers. The experiments were carried out to

establish the relationship between composition - TMAl/III ratio - thickness - relaxation and

to investigate the structural quality of the epitaxial layers. The critical thickness of AlGaN

on AlN was also calculated by different models.

The second section is concentrated on the growth study of novel BAlN materials, in-

cluding single layers and BAlN/AlN heterostructures. The BAlN layers with high boron

content up to 12% were realized. The influences of growth conditions, the structural fea-

tures and optical properties were studied.

The AlN/sapphire templates used for the growth of BAlGaN materials and structures

were provided by the group of Prof. Dupuis from Center for Compound Semiconductors

of Georgia Institute of Technology. The details of the AlN template growth conditions

and optimizations can be found in Ref. [118, 119]. The thickness of the AlN layer for

the templates used in this work were 900 nm. The quality would be characterized and

discussed along with AlGaN layers and MQWs.

42

Figure 15: Al composition of AlGaN layers fully-strained on AlN templates versusTMAl/III ratio. The inset shows the growth rate versus total flow rate of (TMAl+TMG).

3.1 MOVPE growth of AlGaN single layers3.1.1 Control of composition and relaxation

AlGaN single layers were grown on AlN/sapphire templates at 1000 ◦C under 133 mbar.

Hydrogen was used as carrier gas. The composition and strain state (plastic relaxation)

were determined by XRD symmetric and asymmetric reciprocal space mappings (RSMs).

It was verified for one AlGaN sample from RSM measurements along 6 asymmetric ϕ

reflections of the wurtzite lattice [106], that the tilt disorientation could be neglected in

the following XRD spectra analyses. The growth rate was determined from the in-situ

reflectance oscillations for thick layers, and by fitting the Pendellosung fringes of 2θ − ω

scans for the thickness of thin layers. The thickness estimation was moreover confirmed by

STEM analysis.

For the AlGaN thin layers which are fully-strained on AlN templates, Al composition

in the solid phase is almost the same as the TMAl relative concentration in the gas phase

(Fig. 15). Additionally, the AlGaN growth rate, shown in the inset, varies linearly with

the total III elements flow (TMAl+TMGa) while the V/III ratio remains constant, which

indicates that the growth occurs in a mass transport limited regime.

43

Figure 16: (0 0 0 2) 2θ−ω scan with simulation and (1 1 -2 4) reciprocal space mapping for(a): 24-nm thick AlGaN fully strained on AlN template and (b): 350-nm AlGaN layer with55% relaxation on AlN template. Both samples are grown under a fixed TMAl/III ratio of57%.

However, when the thickness of the single AlGaN layer was increased under the same

TMAl/III ratio so that the layer relaxed, it was observed that the average Al content in

the layer decreased. As an example, Fig. 16 shows the 2θ − ω scans and RSMs for the

two samples grown under the same TMAl/III of 57% but having different thicknesses: the

29-nm thick AlGaN is fully-strained on AlN template and shows an Al content of 0.57 (±

0.01), while for the 350-nm thick layer, showing a 55% relaxation, the Al content is only

0.47 (± 0.01).

This composition fluctuation during AlGaN relaxation has generally been related to

composition pulling effect [120–122]. G.B. Stringfellow et al. explained that the excess lat-

tice mismatch energy would perturb the solid composition towards the composition which

44

Figure 17: Al content in the AlGaN single layers plotted as a function of (a) layer thicknessand (b) corresponding layer relaxation for three different TMAl / (TMAl+TMG) ratios.

minimizes mismatch (composition pulling effect) [123]. The smaller adatoms will be in-

corporated preferentially at steps having relative compressive strain, and larger adatoms

under relatively tensile strain [124, 125]. But very few studies have considered AlGaN

layers grown on AlN templates which would be under compressive strain.

In the present investigation, as shown in Fig. 17, a decrease in the Al composition of

AlGaN layers is evidenced when the layer thickness (and hence the layer relaxation) is

increased. For three different TMAl/III ratios in the gas phase to cover the range of 20%

∼ 70% Al, a clear Al content drop can be observed, confirming the composition pulling

effect. In our case, AlGaN is under compressive strain on AlN, so in the initial stage

when the layer is fully strained, AlGaN has a tendency towards higher Al content so as to

minimize mismatch, while for the relaxed layer case, a lower Al content is observed. Since

Ga-N has a smaller bond energy than Al-N, Ga incorporation would be more controlled by

the strain state than Al incorporation [126, 127], which means that Ga atoms are expelled

out for the initial stage under high compressive strain, and Ga incorporation increases when

the layer is relaxed with lateral lattice increasing.

This composition pulling phenomenon should be paid attention to, not only for our

45

application here, but also for all other AlGaN-based devices.

3.1.2 Critical thickness for AlGaN grown on AlN templates

This subsection attempts to predict critical thickness range of AlGaN layers grown on AlN

templates by applying different theoretical models reported in the literature.

• Van der Merwe model (VM)

The theoretic model of calculating critical thickness of epitaxial layers was first pro-

posed by J.H. van der Merwe [128, 129]. He initially formulated a misfit dislocation theory

for critical thickness of strained layers assuming that the crystal would reach thermody-

namic equilibrium and settle into the lowest energy state [128]. In brief, the critical re-

laxation point is where the interfacial energy εI between the film and the substrate for

generating dislocations is equal to the strain energy in the film [6]. The interfacial energy

εI can be expressed by the following equation (for moderate misfit f ≤ 4%):

εI = 9.5 · f · (Gb4π2 ) , (18)

where G is the shear modulus and b is the slip distance (or Burgers vector). The areal strain

energy density associated with a film of thickness h is:

εH = 2G(1 + ν

1 − ν)h f 2 , (19)

where ν is Poisson’s ratio. By setting εI = εH, the critical thickness hc can be obtained:

hc ' (9.58π2 )(

1 − ν1 + ν

)bf, (20)

where a0 is the bulk lattice constant of the substrate.

•Matthews-Blakeslee mdoel (MB)

J.W. Matthews and A.E. Blakeslee have proposed Matthews-Blakeslee model for pre-

dicting the critical thickness via mechanical equilibrium theory instead of minimizing the

total energy of strain and dislocations in a crystal [130].

Figure 18 shows a grown-in threading dislocation in coherent interface (a), critical in-

terface (b) and incoherent interface (c). FD is the tension in the dislocation line and FH

46

Figure 18: Elongation of a grown-in, threading dislocation to form a length LL’ of misfitdislocation line. [6]

is the force exerted on the dislocation line by misfit stress. For the film thickness ha, the

interface is assumed coherent; for the thickness of hb, the interface is critical (FH = FD);

and for hc where FH > FD, the dislocation is elongated in the plane of interface forming a

misfit dislocation (MD) line with length LL’ [6, 130].

Assuming that the elastic constants of the two media A and B are equal, FH and FD are

given as:

FH ' G(1 + ν

1 − ν)bhε

FD 'Gb2

4π(1 − ν)[ln(

hb

) + 1] .(21)

For the critical point: h=hc and ε=f, so the critical thickness can be expressed in the

following equation:

hc ' (bf

)[1

4π(1 + ν)][ln(

hc

b) + 1] . (22)

By this mechanism, the onset of interfacial misfit dislocations is determined by the

mechanical equilibrium of an existing grown-in threading dislocation. It is not expected to

be operational when the misfit is small or the substrate has high quality in which case the

threading dislocation density is low [6].

• People and Bean model (PB)

47

The PB model is similar to the theory of Van der Merwe in a way that the critical thick-

ness is determined by minimizing total energy in the crystal. It assumes that the growing

film is initially free of threading dislocations. The interfacial misfit dislocations will be

generated where the strain energy at a certain thickness is equal to the minimum of dislo-

cation energy [6]. Normally screw threading dislocations are considered since they have

minimum energy density.

The areal energy density associated with an isolated screw dislocation at a distance h

from a free surface is:

εD ' (Gb2

8π√

2a)ln(

hb

) . (23)

The area strain energy density associated with a film of thickness h is:

εH = 2G(1 + ν

1 − ν)h f 2 . (24)

The critical point is set to be when εD = εH:

hc ' (1

16π√

2)(

1 − ν1 + ν

)[b2

a][(

1f 2 )ln(

hc

b)] . (25)

The critical thickness predicted by PB model can be much higher than the real value in

our case, because the AlN template grown on sapphire always has high dislocation density

(more than 109cm−2), which lowers activation energy of relaxation.

• Fischer model

The Fischer model considers the elastic interaction between straight misfit dislocations.

This method provides an equilibrium theory which predicts the critical thickness of strained

layers and describes the strain relief via plastic flow [7]. As shown in Fig. 19, an “image

dislocation” is put outside the crystal to satisfy the free surface boundary conditions.

The equation of predicting critical thickness can be described as:

hc 'bcosλ

2 f[1 +

1 − ν/44π(1 + ν)cos2λ

ln(hc

b)] , (26)

where λ is the angle between the Burgers vector and the interface. Here the dominant

edge-type dislocations are considered, so the value of λ is assumed to be 0.

48

Figure 19: Schematic illustration of the configuration of real and image misfit dislocationsin a strained heteroepitaxial structure proposed by A. Fischer et al. [7]

• Adjusted Griffith model par Griffith model is normally used to describe the brittle

fracture [131]. The AlGaN layers grown on AlN templates are under compressive strain

which will not be relaxed by forming cracks. But the simple relationship between critical

thickness hc and misfit strain ε from Griffith model can be used:

hc ∝1ε2 . (27)

The previous work by M. Abid has succeeded in giving an experimental curve of critical

thickness for AlGaN layers grown on GaN templates [9]. The starting point was the Griffith

model and the intrinsic parameters were adjusted according to the experimental data. The

obtained critical thickness hc and the misfit strain ε can be expressed as the formula of

composition of Al (x):

hc ' 5.3321x−2.051,

ε2 = (3.122x + 3.189(1 − x) − 3.189

3.189)2 .

(28)

In our case, AlyGa1−yN grown on AlN templates has a compressive strain, and the

sqaure of the misfit strain is:

ε2 = (3.122y + 3.189(1 − y) − 3.122

3.122)2 . (29)

49

Figure 20: Critical thickness of AlGaN layers grown on AlN templates calculated by dif-ferent models along with experimental data.

So the critical thickness for AlyGa1−yN grown on AlN templates would be:

hc ' 5.3321[1.02(1 − y)−2.051] ' 5.439(1 − y)−2.051 . (30)

The parameters (Poisson ratio ν, misfit f and lattice a) used in the calculations are sum-

marized in Tab. 6. For AlGaN/AlN system, two plastic relaxation mechanisms have been

proposed in the literature [132]: a-type (edge) treading dislocations (TDs) with Burgers

vector b = 1/3<1 1 -2 0> from the AlN substrate are inclined toward the <1 -1 0 0 > di-

rections when entering the AlGaN film, or mixed type TDs with Burgers vector b = 1/3<1

1 -2 3> glide on {0 -1 1 1} planes bend along the <-2 1 1 0> directions to generate misfit

dislocations. In the AlN templates used, the edge type treading dislocations are dominant

of which the details can be found in the following subsection. So the first process were

considered in the calculations, except that in the PB model the Burger’s vector of screw

threading dislocations b = <0 0 0 1> was used. The curves of critical thickness versus Al

50

Table 6: Parameters used for the estimation of critical thickness.AlN GaN AxGa1−xN

ν 0.203 [106] 0.183 [106] 0.183+0.02xMisfit of AlxGa1−xN/AlN 0.0215(1-x)

Lattice a 3.122 3.189 3.189-0.067x

content calculated based on different models are shown in Fig. 20

Although the critical thickness predicted by different models has a large dispersion,

the calculations give a range of the possibilities. If the experimental data are located in

the figure, the critical thickness is more or less close to the curve of adjusted Griffith model

compared with the others, which can be used as a reference for a rough estimation of critical

thickness.

It should be noted that the process of misfit strain relaxation of AlGaN on AlN is com-

plex with various different mechanisms. For example, if the substrate has very low disloca-

tion density (<105 cm−2), AlGaN can not relax effectively by the sparseness of pre-existing

dislocations and alternative pathways of elastic surface roughening or misfit dislocating

may take over [133]. Surface roughening can relax partially the misfit strain by purely

elastic deformation of the film and substrate. At large misfit, the surface becomes rough

enabling easy nucleation of dislocations. But this mechanism is kinetically unfavorable at

low misfit, because the strain already relaxes by nucleation or movement of dislocations

before the surface has enough strain to roughen [134].

3.1.3 Estimation of threading dislocation densities by XRD

[0 0 0 1] oriented III-nitrides normally contain three types of threading dislocations (TDs):

a-type (edge) with b = 1/3<1 1 -2 0>, c-type (screw) with b = <0 0 0 1> and a+c type

(mixed) with b = 1/3<1 1 -2 3>. Each dislocation type accommodates a lattice distortion.

Edge TDs lead to a lattice twist, screw TDs lead to a lattice tilt and mixed TDs contributes

to both.

In order to study the structural quality of substrates, epitaxial layers or devices, XRD

51

is an effective and non-destructive method, which is not only for determination of com-

position but can also provide information on threading dislocation densities in the layers

including both edge and screw types (components). ω-scans of symmetric planes can mea-

sure quantitatively the lattice tilt from screw TDs or screw component of mixed TDs. Twist

caused by edge TDs or edge component of mixed TDs should be measured by skew sym-

metric ω-scans, because asymmetric ω scans are sensitive only to tilt [106]. FWHM (full

width at half maximum) of a skew symmetric rocking curve is broadened by both tilt and

in-plane twist, so a series of skew symmetric rocking curves for different planes (different

inclination angle χ) should be done to separate tilt angle and twist angle, as expressed in

Eq. 31, which is particularly useful for high defective layers such as AlN [106, 135]:

β2 = (β2twist − β

2tilt)sin2χ + β2

tilt , (31)

where β is FWHM angle of the skew symmetric rocking curve, βtilt and βtwist are the tilt and

twist spread, and χ is inclination angle between the reciprocal lattice vector and the c axis.

Then, the densities of screw TDs and edge TDs may be estimated by Eq. 32

Nscrew = β2tilt/(4.35 × b2

c)

Nedge = β2twist/(4.35 × b2

a) ,(32)

where the Burgers vector of screw type TDs (bc) and edge type TDs (ba) can be seen equal

to the lattice c and lattice a of the material.

Skew symmetric rocking curves have been done for AlN/sapphire template, 29-nm

Al0.57Ga0.43N layer and 630-nm Al0.58Ga0.42N layer in order to investigate the quality of the

substrates and epitaxial layers. bc is 0.4982 nm for AlN and 0.5067 nm for Al0.58Ga0.42N,

and ba is 0.3112 nm for AlN and 0.3144 nm for Al0.58Ga0.42N. The measured data to iden-

tify tilt and twist angle are shown in Fig. 21, and TD densities are summarized in Tab.

7

The results confirm that the thin and thick AlGaN layers are grown without generating

extra threading dislocations that would propagate into MQWs. Some annihilation of dislo-

cations originating from the AlN template may even happen during the growth of the thick

52

Figure 21: Skew symmetric scans for of AlN template, 29-nm Al0.57Ga0.43N and 630-nmAl0.58Ga0.42N.

Table 7: Threading dislocation densities of AlN templates and AlGaN layers determinedby XRD rocking curves (FWHM determination and linear fitting lead to an estimated errorof 15%).

Type of TDs AlN 29-nm Al0.57Ga0.43N 630-nm Al0.58Ga0.42NScrew TDs (cm−2) ∼ 9.0 × 109 ∼ 6.7 × 109 ∼ 5.6 × 109

Edge TDs (cm−2) ∼ 2.7 × 1011 ∼ 1.6 × 1011 ∼ 1.5 × 1011

53

buffer layer, which could explain the reduced FWHM of the rocking curves.

3.2 MOVPE growth of BAlN with high boron content

Compared with conventional AlGaInN system, Boron containing III nitrides are quite new

in this family. They are suitable for DUV applications because of their wide bandgaps,

and they can bring additional freedom in engineering the bandgap, lattice constant and

refractive index of multi-layered devices. Most of the studies concentrate on BAlN single

layers with only 1∼2% boron. High boron containing layers grown by MOVPE have not

been progressed a lot. This section describes the growth and characterizations of BAlN

layers or heterostructures with high boron incorporation. The properties of the materials

are studied and discussed.

3.2.1 BAlN/AlN grown at 1000 ◦C

The fabrication of BAlN/Al(Ga)N heterostructures is an important issue which needs to

be further developed no matter for BAlGaN-based MQWs or for DBRs. BAlN/Al(Ga)N

heterostructures were grown at 1000 ◦C by both flow-modulate epitaxy (FME) and con-

ventional continuous method. The experiments were performed in T-shape reactor at 133

mbar by using hydrogen as carrier gas. In this subsection the samples by FME method

are analyzed in details, and the results of this part can also be found in Ref. [136]. The

structural quality of the heterostructures grown by continuous method is investigated in the

subsection 5.4.1 for DBRs.

The samples consist of 5-period AlN/BAlN layers (25 nm / 32 nm) on two types of

substrates: 1 µm AlN templates on sapphire and 3 µm GaN templates on sapphire. The

AlN templates are appropriate substrates for deep UV applications while GaN templates

were used as reference. The temperature was maintained at 1000 ◦C during the growth and

TEB/III molar ratio in the vapor phase was 39% in order to have high boron incorporation.

Flow-modulate epitaxy (FME) was applied during the growth of BAlN layers in order to

enhance the migration of B and Al atoms and also to suppress parasitic reactions [72,

54

Figure 22: Schematic of precursors feeding sequence for BAlN and AlN.

75, 81]. 2 s supply of metalorganics and 1 s supply of NH3 were alternatively run into

the reactor without interruption. The AlN layers were grown in a continuous way. The

schematic of precursor feeding sequence for BAlN and AlN is presented in Fig. 22.

3.2.1.1 Structural characterizations

The boron concentration in the BAlN layers along growth direction was evaluated by SIMS

profile, as shown in Fig. 23(a). It was clear that B profile varies anti-phase with Al, which

indicates that boron atoms substitute aluminium atoms on the III sites of the lattice to form

an alloy. 5-period AlN/BAlN structure exhibits good uniformity except that the first AlN

layer has lower AlN intensity which is due to some Ga contamination from the sample

holder and the reactor [137]. It should be pointed out that the boron signal cannot be

zero when it is sputtered into AlN layer considering SIMS detection limit when thin layers

are analyzed. In order to calibrate SIMS signal for quantitative measurements of boron

content in the layer, boron implanted AlN sample was used as a reference. The boron

content distribution along the growth direction is presented in Fig. 23(b). Under our growth

conditions, 11% (± 0.6%) boron incorporation has been obtained.

55

Figure 23: (a) SIMS elemental concentration depth profiles of B and Al for the samplegrown on GaN template; (b) Boron content in solid layers calculated from SIMS by usingboron implanted AlN sample as reference.

56

Figure 24: (a) STEM images (bright field) of 5-period AlN/BAlN heterostructure andcolumns are clearly observed in the structure; (b) HAADF-STEM image to show bettercontrast of BAlN and AlN layers; (c) high magnification of the zone where the 1st BAlNlayer starts to grow.

In order to investigate structural quality of this heterostructure and also crystalline char-

acteristics, the cross-section STEM was performed along 〈1 1 -2 0〉 zone axis. As shown

in Fig. 24(a), the bright-field STEM image shows that the AlN/BAlN heterostructure has

columnar polycrystalline features, such as the part in the rectangle box. By looking into the

higher magnification image of the interface between 1st AlN and 1st BAlN in Fig. 24(c),

it is clear that the 1st AlN layer is still monocrystalline. When BAlN (with 11%) boron

growth starts, the lattice is oriented along c-axis for around 5 nm, and then the tilt as large

as 60◦ can be observed that means the structure tends to be polycrystalline and colum-

nar growth starts. Better contrast of AlN and BAlN layers can be observed by Z-contrast

HAADF-STEM image shown in Fig. 24(b), where layers with higher brightness represent

AlN layers and darker layers represent BAlN. The surface roughness height caused by this

columnar feature is around 10∼13 nm from STEM images.

The polycrystalline feature has been confirmed by HR-XRD results. In 2θ-ω scans

57

Figure 25: HR-XRD 2θ-ω scans of 5-period AlN/BAlN heterostructure grown on (a) GaNtemplate and (b) AlN template.

shown in Fig. 25, a peak related to AlN (0 0 0 2) was located at 36.02 ◦. Another peak

at 37.98 ◦ should correspond to AlN (1 -1 0 1). Combined with STEM results, it can be

explained that the first AlN layer was monocrystalline along c-axis. After the BAlN starts

to growth, the structure becomes polycrystalline and epitaxial AlN layers grown afterwards

also have lattice tilt so that x-ray diffraction signal from other facets arises. BAlN XRD

peak is absent due to its polycrystalline feature especially with high boron incorporation.

This AlN/BAlN heterostructure has total thickness of 310 nm. The morphology was

examined by AFM shown in Fig. 26. The surface was covered by columnar crystallites

58

Figure 26: (a) SEM and (b) AFM images of 5-period AlN/BAlN heterostructure (310 nmfor total thickness).

confirming the STEM observations. The AFM scan area is 5 µm × 5 µm square. The

structure has root-mean-square (RMS) of 3.3 nm. The average height of these columnar

crystallites is around 10 nm which is in a good agreement with estimations from cross-

section STEM images.

The polycrystalline feature is caused by the short diffusion length of boron atoms, which

would challenge the applications of this material. From STEM image, it was observed that

the monocrystalline critical thickness for BAlN with 11% boron is around 5 nm, above

which the polycrystalline growth occurs. Meanwhile, the monocrystalline critical thickness

is around 500 nm for BAlN of 2% boron as reported in the literature [72]. The more boron

is incorporated, the smaller monocrystalline thickness of BAlN is. Therefore, for different

applications, a compromise can be achieved between thickness and boron composition. For

example, for deep UV DBRs, boron incorporation no more than 5% is enough to achieve

high refractive index contrast theoretically [63, 64]. So it can be an option to reduce the

boron content and maintain BAlN layers (30∼40 nm) monocrystalline. For ultra-thin layers

such as MQWs or strain engineering superlattices, higher boron incorporation can be used

allowing a large design freedom and it can still be kept as monocrystalline for its thin

thickness (below 10 nm).

59

3.2.1.2 Optical characterizations

In order to study optical properties of this heterostructure containing high boron, Cathodo-

luminescence (CL) and transmission spectra have been performed for the sample grown on

AlN template. As shown in Fig. 27, a well-defined emission peak at 262 nm emerges when

the excitation power is above 7 keV. Meanwhile, as shown in Fig. 28, the optical absorption

occurred at 260 nm which agrees well with CL results. For the wavelengths longer than 290

nm, the oscillation level of transmission fraction keeps constant, which implies that there

is very less absorption by defects in this region. The results obtained exhibit potentials to

apply this material and structure for UV and deep UV devices.

It is noted that CL emission peak of AlN has not been detected due to limitation of our

detector below 210 nm. In addition, since emission wavelength of BAlN is at 260 nm, the

emission around 200 nm of AlN would be absorbed by the BAlN layers. As shown in Fig.

28 of transmission curve, there is a transmission drop at 260 nm, and below 260 nm, there

is a large absorption.

Until now, few results have been reported on optical properties of BAlN material, es-

pecially for high boron containing layers. Theoretical calculations show that BAlN has

strong bowing parameter (5.45 eV) [86], and for BGaN material the value of 9.2 eV has

been experimentally obtained [74]. Based on this prediction, our BAlN layers should give

an emission at around 225 nm. Here a significant redshift of wavelength was observed

which might be due to high concentration of carbon impurities incorporated from TEB

precursors (carbon density about 2×1019 cm−3 by SIMS). The optical properties of BAlN

requires more experimental investigations.

3.2.2 BAlN grown at low temperature with annealing

In the last subsection, the high boron content of BAlN grown at 1000 ◦ was determined

by SIMS. No XRD peak related to the BAlN material was found. High boron content was

also demonstrated in the literature [69], but the related XRD peak intensity was very weak.

The objective of this subsection is to study the influences of growth conditions, especially

60

Figure 27: Cathodoluminescence spectra at 77 K of 5-period AlN/BAlN heterostructuregrown on AlN template.

Figure 28: Transmission spectrum at room temperature of 5-period AlN/BAlN heterostruc-ture grown on AlN template.

61

Figure 29: Schematic of growth procedure.

the growth temperatures, in order to improve the crystallinity to have clear XRD peaks of

BAlN layers [138].

3.2.2.1 Influence of growth temperature

BAlN layers were grown at 133 mbar by using hydrogen as carrier gas. The growth was

performed on two types of substrates: 1 µm AlN templates on sapphire and 3 µm GaN tem-

plates on sapphire. As shown in Fig. 29, the substrates were annealed in hydrogen at 1040

◦C prior to growth. Thin BAlN layers were grown in a continuous way at low temperatures,

which were varied from 650 to 800 ◦C to study the influences. Then the temperature was

ramped up to 1020 ◦C and the samples were annealed for 5 min before cooling down. A

large flow rate of NH3 (2.3 slm/min) was used due to inefficient decomposition of NH3 at

low temperature. Under high TEB/III ratio of 39%, as shown in Fig. 30(a), BAlN single

layer grown at 650 ◦C with 20 nm thickness demonstrates an X-ray diffraction peak at 36.38

◦ (± 0.17 ◦), which indicates that the layer has a smaller lattice c than AlN due to boron

substituting Al atoms in the crystal structure. It is assumed that the layer is fully relaxed

considering the large lattice mismatch between BAlN layer and GaN template. In this case,

the c-lattice constant is 4.935 Å (± 0.022 Å), which corresponds to boron composition of

5.6% (± 2.8%) by applying Vegard’s Law. The complete relaxation of the layer can be

62

Figure 30: (a) HR-XRD 2θ-ω scan of 20 nm BAlN layers grown on GaN template at 650under TEB/III=39%; (b) shows the influence of growth temperatures which was variedfrom 650 ◦C to 800 ◦C .

confirmed in Fig. 31; when TEB flow is stopped for growing AlN layer, the 2θ-ω peak of

the layer is located at 36.02 ◦ corresponding to completed-relaxed AlN layer.

When the deposition temperature is increased, shown in Fig. 30(b), the BAlN peak is

weakened, and then disappears when it is increased to 800◦C. It indicates that under this

high TEB/III ratio the crystallinity is worse when the layer is grown at higher temperature.

Low temperature growth can alleviate the B-rich phase poisoning issue under high TEB/III

ratio. So the following results in this subsection are all referred to the growth at 650 ◦C.

3.2.2.2 Influence of TEB/III ratio

In order to have a different amount of boron incorporation, the flow rate of TMAl was

maintained constant while flow rate of TEB was varied to have TEB/III molar ratio in

the vapor phase of 0%, 9%, 17%, and 39%, respectively. As shown in Fig. 31(a), when

TEB/III ratio is increased from 0 to 39%, the peak of the layer shifts gradually towards

63

Figure 31: HR-XRD 2θ-ω scan of 20 nm BAlN layers grown on GaN templates by contin-uous method under different TEB/III ratio.

greater diffraction angles (0 to 5.6% boron). The peak is broadened with fringes missing

due to polycrystalline nature of BAlN, which will be discussed later. Figure 31(b) shows

that boron content in solid phase has almost a linear relationship with TEB/III ratio in the

gas phase. The morphology change was studied by AFM in Fig. 32. As observed, without

boron incorporation, 40 nm AlN layer on GaN template exhibits normal V-defects and

cracks generated by lattice mismatch. The flat area be-tween defects has surface roughness

around 0.7 nm. When boron was incorporated, the surface became rough, and cracks and

v-defects disappeared. The surface is covered by small crystallites caused by columnar

growth of BAlN, and this columnar feature would be clearly observed by STEM results.

Surface roughness was around 1.2 nm for TEB/III=39% (∼5.6% boron in solid phase).

64

Figure 32: AFM images of 40 nm BAlN layers grown on GaN templates under TEB/IIIratio of 0% and 39%.

3.2.2.3 BAlN on AlN by FME

For BAlN layers grown on AlN templates by continuous flow, the XRD peak of BAlN layer

was too weak and it cannot be distinguished from noise. So FME method was applied to

improve the crystalline quality. The same feeding sequence in Fig. 22 was used. BAlN

single layers with clear defined X-ray diffraction peaks were also achieved on AlN tem-

plates which are appropriate substrates for deep UV devices, as well as on GaN templates,

in order to distiguish the XRD peak of BAlN from the substrate peak. As shown in Fig. 33,

clear XRD peaks can be identified at 36.59 ◦ (± 0.20◦) on both GaN and AlN templates.

Lattice c is 4.908 Å (± 0.025 Å) and the corresponding boron content is 9% (± 3.2%).

Since BAlN peak is very close to AlN template peak, the deconvolution of substrate peak

and layer peak is shown in the inset figure.

The concentration calculated by XRD has a large error range because of the broadness

of the BAlN diffraction peak and uncertainty of the lattice parameters and strain. These

all influence calculation of the composition from X-ray diffraction peak positions, espe-

cially for the layer with high boron content. So the boron incorporation into the layer was

also analysed by SIMS profile along the growth direction in Fig. 34. The Al signal is de-

creased in the BAlN layer compared with the signal of the template indicating that boron

65

Figure 33: HR-XRD of 70 nm BAlN on (a) GaN template and (b) AlN template by FMEgrowth (TEB/III=39%). Inset figures show the smoothing and deconvolution of two peaks.

66

Figure 34: SIMS elemental concentration depth profiles of B and Al for the sample grownon AlN template; inset shows the boron concentration obtained by using boron-implantedAlN as calibration sample.

atoms were substitutionally incorporated into AlN lattice. Boron has a uniform distribution

along the growth direction. The concentration of boron can be calculated based on atomic

concentration obtained by SIMS with a boron-implanted AlN sample as a reference. The

concentration of boron calculated from the SIMS signal is 12% with 0.6% error (inset of

Fig. 34), which agrees with the composition range given by XRD diffraction peak positions

(6% ∼ 12%). STEM characterizations were also performed to study the crystallinity of

the layer. Cross-section image in Fig.35 shows that BAlN layer has a columnar crystalline

growth feature. At the beginning of BAlN growth, the lattice is still oriented along c-axis

(the zone A). After around 10 nm, the crystals start to disorient and form columns (zone

B). The monocrystalline thickness is higher than the one grown at 1000 ◦ by FME. The

lattice of columnar crystals can be clearly observed in Fig. 35 (b) which is the top part of

the layer. This columnar growth feature would lead to surface roughness, as observed in

AFM analysis.

Fig. 36 presents HAADF-STEM image of the BAlN layer and diffraction pattern after

67

Figure 35: (a) Cross section STEM image (bright field) of 75 nm thick BAlN layers con-taining 12% boron along the [1 1 -2 0] zone axis. Zone A has lattice oriented along c-axisand Zone B has columnar feature; (b) higher magnification image for the top part of thelayer; (c) higher magnification image for the film/substrate interface.

Figure 36: Cross section High-angle Annular Dark Field Scanning Transmission Mi-croscopy (HAADF-STEM) image of 75 nm BAlN layer containing 12% boron; inset showsdiffraction pattern after Fast Fourier Transform (FFT).

68

fast Fourier transform (FFT). It exhibits mainly the typical pattern of the wurtzite crystal

along the <1 1 -2 0> zone axis, even though the diffraction spots are elongated due to the

different tilts of columnar polycrystals.

69

CHAPTER 4

DEEP UV AlGaN MQWS: DESIGN, GROWTH ANDCHARACTERIZATIONS

This chapter is addressed to the AlGaN MQWs emitting at 280 nm: design, growth and

characterizations. The simulation and the design of MQWs to enhance surface emission

were done by the partners in Institut Pascal - Universite Blaise-Pascal (LASMEA, Aubiere,

France), and the basic principles are described in the first section. The structural and optical

properties of 4-, 10- and 20-period MQWs are studied in the second and third sections.

The experimental results confirmed the preserved oscillator strength of surface emission

as designed. Besides, the analysis of typical defects in QW samples is presented in the

last section. Their influences on the structural and optical properties of the MQWs are

investigated.

4.1 AlGaN MQW design for enhanced TE (E⊥c) emission4.1.1 Principles of AlGaN band structure calculation

III nitrides (AlGaN) has a direct wide bandgap over the entire composition range. The

valence band (VB) splits into three bands due to crystal-field splitting and spin-orbit in-

teraction. The schematic of band ordering is shown in Fig. 37. The energies of three

bands (Γ9, Γ7+ and Γ7−) are related to crystal-field split-off energy ∆cr and spin-orbit split-

off energy ∆so. The values of ∆cr and ∆so can be determined by measuring the interband

optical-transition energies EAV - EB

V and EAV - EC

V [139].

∆cr of GaN is positive, so the order of VBs is EAV (Γ9), EB

V (Γ7+) and ECV (Γ7−). On

the other hand, ∆cr of AlN is negative, and EBV (Γ7+) becomes the top valence band. So

in AlGaN alloys, the order of the two upper valence bands is modified along with the

increase of Al content, as shown in Fig. 38. Apart from the modification of the electric

band structure, alloying also leads to a modification of the optical polarization properties

of the interband transitions which influences the performance of the devices. In GaN, the

70

Figure 37: The band-edge energies with and without spin-orbit interaction.

upper valence band corresponds to the heavy hold band of Γ9 symmetry, while the second

one corresponds to the light hole band of symmetry Γ7+. For an aluminum composition

typically higher than 10%, the energy order of these two bands reverses and the band of

Γ7+ symmetry becomes the upper valence band [88–92]. This energy crossover between

AlN and GaN is also accompanied by a switching of the valence band state symmetry.

The topmost Γ7+ valence band is governed by pz-like state. The following bands (Γ9 and

lower Γ7+) are governed by px and py-like states. So, if the three valence bands are labeled

according to their zone center wavefunctions, the appropriate notation for AlN but also

for AlGaN with high Al composition becomes CH (crystal field split-off band), HH (heavy

hole band) and LH (light hole band) from top to bottom. The oscillator strength between the

conduction band and the upper Γ7+ (split-off hole, CH) valence band, which becomes the

fundamental transition, is highly reduced with increasing Al fraction for the configuration

where the electric field is perpendicular to the c axis [88–92].

The band alignment of AlGaN is shown in Fig. 38. The band energies of AlGaN

ternary alloy can be deduced from band energies of AlN and GaN with considering bowing

71

Figure 38: Band alignment of GaN/AlGaN structure [4].

parameter [4]:

Ec(AlGaN) =Eg(GaN) + x∆Ec − ηbx(1 − x)

EAv (AlGaN) = − x(∆Ev + ∆Γ79(AlN)) + (1 − η)bx(1 − x)

EBv (AlGaN) = − x(∆Ev − ∆Γ97(GaN)) + (1 − η)bx(1 − x) − ∆Γ97(GaN)

ECv (AlGaN) = − x(∆Ev + ∆Γ79(AlN) + ∆Γ

(S O)97 (AlN) − ∆Γ

(S O)97 (GaN))

+ (1 − η)bx(1 − x) − ∆Γ(S O)97 (GaN) ,

(33)

where ∆Ev and ∆Ec are CB and VB offset. ∆Γ97(GaN), ∆Γ(S O)97 (GaN), ∆Γ97(AlN) and

∆Γ(S O)97 (AlN) are positive parameters of the band energy differences. η and 1−η are fraction

of bowing b on the conduction band and valence band.

Besides the composition, the strain can also affect the valence band states and the se-

lection rules [139–143]. Based on the work of Chuang and Chang [139], the strained effect

on VB energies at Γ point can be considered by adding the terms θε which represents the

energy change due to uniaxial strain and λε which represents the energy change due to

72

biaxial strain:

EAV(Γ9) = ∆cr +

13

∆so + θε + λε

EBV(Γ7+) =

12

(∆cr −13

∆so + θε) + λε +

√√√(∆cr −

13

∆so + θε

2)2 +

29

∆2so

ECV (Γ7−) =

12

(∆cr −13

∆so + θε) + λε −

√√√(∆cr −

13

∆so + θε

2)2 +

29

∆2so ,

(34)

where the uniaxial term and biaxial term can be expressed as the function of the deforma-

tion potentials D1−4 of the material and strain terms εxx, εyy and εzz:

θε = D3εzz + D4(εxx + εyy)

λε = D1εzz + D2(εxx + εyy) .(35)

The strain terms can be calculated based the lattice parameters ”a” before and after defor-

mation and elastic stress constants ”C” as follows:

εxx = εyy =astrained − aoriginal

aoriginal

εzz = −2C13

C33εxx .

(36)

4.1.2 Design of AlGaN MQWs

The amplitude of the oscillator strength for TE polarization can be restored by imposing

some strain in the AlGaN QW. The strain can be due to the lattice mismatch between

barriers and wells which have different Al content. Thus, the barrier composition can

be chosen to provide sufficient compressive strain in the wells, and to enhance the TE-

polarized optical transition.

The calculations of band structures and design of AlGaN MQWs were performed by

LASMEA (Aubiere, France). The band structure and the optical interband matrix elements

of AlGaN compounds were carried out by using k ·p formalism for strained wurtzite semi-

conductors taking into account both valence band mixing due to crystal field and spin orbit

effects and strain effect [140], as described in the last subsection. For the design of the Al-

GaN/AlGaN MQW structure, envelop function simulations taking into account strain and

73

built-in electric fields have been performed. Considering the large scatter of valence-band

offset (VBO) values reported in the literature (0.3 eV to 0.7 eV for GaN/AlN) [144–148],

the experimental value of 0.5 eV determined by Baur et al. [144] for the VBO of GaN/AlN

interface in the absence of strain is used. This value has already been tested on GaN/AlGaN

QWs with a good agreement between calculations and experimental data [4]. The band dia-

gram for the AlGaN/AlGaN heterostructure is then constructed by distributing the bandgap

bowing on the conduction and valence bands of each alloy and by adding the strain effects

induced by the lattice mismatch between two materials [4, 89, 139, 149]. The relative oscil-

lator strength modeling is calculated by evaluating the matrix element between conduction

and valence bands.

To achieve emission at a wavelength of around 280 nm, the Al composition of the well

was chosen to be xAl = 0.37 and the well thickness was fixed to 1.7 nm in the calculations.

The thickness of barriers was fixed to 10 nm. The calculated values of the relative oscillator

strengths for TE transition are displayed in Fig. 39 as a function of the Al composition in

the barriers. The uppermost valence bands of wurtzite nitrides are formed out of p orbitals

with wave functions combining |X〉, |Y〉 and |Z〉 symmetries. The anisotropic strain mixes

these valence band states and the polarization properties of the interband transitions are

thus modified. When the biaxial stress increases with the increase of Al content in the

barriers, the band-to-band oscillator strength of the fundamental transition involving Γ7-

valence band (crystal field split-off hole band (CH)) increases up to 0.5, which corresponds

to the value of the oscillator strength of the transition involving the Γ9-valence band (heavy

hole). The CH valence band is no longer purely governed by pz states but arises from a

mixing between px, py and pz states and is therefore not forbidden for TE polarization. The

compressive strain increases the weight of px, py-like states at the expense of pz-like states.

So, the optimal Al content in barriers is designed to be 0.57 (±0.01), for which the strain

(-0.5%) introduced in the wells is sufficient to enhance TE-polarized optical transition and

therefore surface emission. The above calculations assume no strain in the AlGaN barrier

74

Figure 39: Relative oscillator strengths for the optical transitions between the valence bands(Γ7 (CH) and Γ9 (HH)) and conduction band (CB) in an AlGaN/AlGaN quantum well asa function of the Al composition in the barriers, with Al content in the well fixed to xAl

= 0.37. The corresponding strain in the well is also reported in the top axis. Calculationsconsider that the barriers are strain-free and QWs are fully-strained on AlGaN barriers.

material.

The strong electric-field in AlGaN quantum wells due to both piezoelectric and spon-

taneous polarization in this structure is around 1.15 MV·cm−1 and tends to separate the

electrons and holes. To solve this problem, ultra-thin wells below 2 nm must be considered

so that the oscillator strength obtained is not counteracted by the QCSE.

Practically, in order to release excess strain in the barriers, it is necessary to start the

growth of the active region from a pseudo-substrate with barrier lattice mismatch as small

as possible. AlN templates grown on c-Al2O3 wafers are appropriate substrates for the

growth of Al-rich AlGaN MQW structures, however, AlN shows a lattice mismatch of 1%

with Al0.57Ga0.43N barriers. Hence, a thick, relaxed Al0.57Ga0.43N layer has to be inserted

before the MQW growth, acting as a latticed-matched buffer. Fig. 40 presents the final

design of the structure.

75

Figure 40: Schematic of MQW structure for emission at 280 nm

4.2 Growth and characterizations of MQWs

AlGaN MQW structures were realized based on the design. Detailed characterizations

relating to the determination of compositions and optical properties are described in this

section.

4.2.1 Structural characterizations

The quality of the relaxed AlGaN buffer is analyzed in Section 3.1.3. The MQW sample

exhibits 2D morphology, and the root-mean square (RMS) surface roughness is 0.45 nm

for 1×1 µm2 scan. Figure 41 shows the corresponding 2θ-ω scan of the sample. The

diffraction pattern is dominated by the strong peaks related to the AlN template and the

AlGaN relaxed buffer. Satellite diffraction peaks (SL) associated with the quantum wells

can also be observed indicating a good quality of MQWs. The RSM in the inset shows

the AlN template spot as well as the broader AlGaN buffer spot (and barriers). The signal

from the thin quantum wells is too weak to be observed in RSM. From the symmetric 2θ-

ω scan and asymmetric RSM, the composition and relaxation degree of the buffer were

76

estimated: the average Al composition of buffer is ∼0.58 (±0.01) and the buffer has 70%

relaxation. The composition and thickness of the wells could not be estimated accurately by

just fitting 2θ-ω scan considering that the simulation of multi-layered structure is influenced

by thickness and composition for each layer simultaneously. Further characterization using

STEM and EDX is necessary in order to obtain more information about the quantum wells.

HAADF-STEM cross-section images of MQWs and upper part of the relaxed buffer

are shown in Fig. 42(a) and 42(b). The barrier thickness is measured to be 10 ∼ 11 nm

from intensity profiles and the well thickness is of 1.6 ∼ 1.8 nm. It can be observed that

after each well there is an Al-rich layer which is presumably caused by a switch between

precursors that can be optimized for the planned future work.

The average composition of barriers could be determined from EDX quantitative anal-

ysis and was found to be 0.57 (± 0.015). The k-factors used for the EDX quantification

have been calibrated using thick AlN and GaN layers layers epitaxially grown on a silicon

substrate. All calibration samples were prepared by FIB (the thickness of the slices is com-

prised between 60 to 80 nm). The systematic control of the stoichiometry (ratio between

the III elements and the nitrogen content measured) ensures to be the right conditions for

quantitative analysis (with accurate k-factors). The accuracy of the EDX analysis is esti-

mated to be 1% (except with nitrogen where the precision is rather ± 2%). The results of

the quantitative analysis (composition in atomic %) does not vary by more than ± 1% when

the slice thickness varies in the range from 60 nm to 80 nm.

Since the spatial resolution of the EDX analysis is typically of 2 nm, high-resolution

HAADF-STEM images have been used to estimate the composition of the thin quantum

wells following a procedure proposed in [110]. The electron beam scattered intensity for

each point M of the background-subtracted HAADF STEM image can be expressed as:

I(M) = d(M) × K ×∑

i

xiZαi , (37)

where d(M) is the STEM slice thickness at point M, K is a proportionality factor depending

on the number of atoms per unit volume in the lattice, the label i corresponds to atom i, xi

77

Figure 41: (0 0 0 2) 2θ-ω scan for 4 quantum wells grown on a relaxed buffer on AlNtemplate and the simulation of the structure which used values obtained by XRD, STEMand EDX analyses. The RSM of (1 1 -2 4) reflection is shown in the inset.

is the relative proportion of atom i (at point M), Zi is its atomic number, and the value of

power α is typically close to 2.

For the AlxGa1−xN layers (x=xAl), Eq. 37 can be written as:

Ix(M)dx(M)

=K2× ZGa

α[a(α) − b(α) × x] , (38)

where a(α) = 1 + ZNα/ZGa

α, and b(α) = 1 − ZAlα/ZGa

α.

In order to use Eq. 38, the background-subtracted intensity Ix(M) is corrected from

slight variations in the STEM slice thickness variations (dx(M)) by comparing the inten-

sity of the HAADF-STEM images at points where the EDX quantitative analysis predicts

a nearly constant Al composition (in the AlGaN buffer, in the barriers). The thickness vari-

ation is extrapolated from these points for the whole HAADF image. A slight, regular and

nearly linear variation of the thickness is observed, except at the very top surface of the

epitaxial structure where a sharper variation of the slice thickness seems to occur. This

region is therefore discarded from the analysis.

Then, the value of power α and of the proportionality constant K can be retrieved from

78

Figure 42: (a) Cross-section High-angle Annular Dark Field Scanning Transmission Elec-tron Microscopy (HAADF-STEM) images taken along <1 1 -2 0> zone axis for MQWsand buffer layer; (b) High magnification of HAADF-STEM images; (c) Al compositionmap obtained from (b).

points in the AlGaN buffer layer and in the AlN template, where xAl is given by the EDX

quantitative analysis. It is moreover assumed that K is constant for the range of materials

considered. Given the values of ZGa = 31, ZAl = 13, ZN = 7, and α lying in the range of

1.6 to 2 [110], it is found that the best consistency with the experimental values in AlGaN

buffer and AlN template are obtained for α = 2 in our case.

The complete analysis results in a chemical mapping, as shown in Fig. 42(c). The aver-

age Al content in the barriers is in accord with EDX value (xAl ∼ 0.57) and the average Al

content in the wells is estimated to be xAl ∼ 0.38 (± 0.015). The thickness and composition

values obtained from Figs. 42(a)-42(c) fit well the XRD experimental data shown in Fig.

79

41.

It is noticed that the composition of 4th well in the mapping is abnormally high (xAl =

∼0.50), and so as the last barrier (xAl = ∼ 0.67) which is not consistent with EDX value. It

might be caused by the fact that the thickness variation near the top surface deviates from

the linear extrapolation and hence leads to the inaccuracy of composition estimation from

Z-contrast intensity.

4.2.2 Optical characterizations

Cathodoluminescence (CL) spectra at 77 K under different excitation energy are shown in

Fig. 43(a). Under a low excitation power of 3 keV corresponding to a penetration depth of

the electron beam of typically 30 nm, a single emission peak from wells at 286 nm indicates

that the carriers are mostly confined in the wells. When the excitation power is increased to

10 keV and the penetration depth of the excitation beam reaches 230 nm, a luminescence

signal at 262 nm appears in addition to the emission of the wells, which is attributed to the

barriers and buffer layer. The emission at 262 nm corresponds to a bandgap energy of 4.73

eV, which is in agreement with the experimental composition of the barriers and buffer layer

(xAl ∼ 0.57). At room temperature the increase of the barrier luminescence with respect to

77K is attributed to the thermal activation of carriers in the AlGaN layers (barriers or buffer

layers). In the case of an optical in-well pumping (excitation at 266 nm) as displayed in

Fig. 43(b), the laser beam is mainly absorbed in the QWs and not in the barriers. Thus,

only the luminescence of the wells is observed. However, it is found that both classical

photoluminescence and cathodoluminescence provide the same QW emission line. The

linewidth is 9.5 nm for PL at 77 K and 11.9 nm for PL at 300 K.

Figure 43(c) shows the transmission measurements (E-field ⊥ c configuration) at 77 K

together with numerical simulations based on transfer matrix formalism. The experimen-

tal spectrum reveals the absorption edge of the barriers at 260 nm, while a 10% drop of

transmission is observed at 281 nm due to absorption in the wells. It is worth noting that

calculations fit the experimental results in a satisfying way. The absorption coefficients

80

Figure 43: (a) Cathodoluminescence (CL) spectra at 77 K (and at 300 K in the inset) fortwo different values of excitation power; (b) Photoluminescence (PL) at 77 K and at 300 Kunder excitation of 266 nm; (c) transmission measurements and transfer-matrix simulationof MQWs together with absorption coefficients (αwell, αbarrier, bu f f er used in the simulation).

81

used in the simulation are also displayed in Fig. 43(c) for both barriers and wells. A weak

absorption with linear energy dependence has been added in the barriers and the buffer

in order to reproduce the overall decrease of the transmission signal. The latter might be

caused by defects originated from the AlN template. The absorption coefficient in the wells

is found to be as high as 3×105 cm−1, which implies that the oscillator strength is preserved

in the QWs despite the high aluminium composition.

It is noted that the splitting between Γ7CH-CB (fundamental) and Γ9-CB transitions is

evaluated to be equal to 32 meV. By considering the AlGaN emission broadening which is

due to intrinsic alloy disorder and extrinsic inhomogeneities such as QW thickness fluctu-

ations, it appears that the Γ7CH-CB and Γ9-CB transitions lie in the same energy range. The

energy difference between these two transitions depends on several parameters (band off-

set, deformation potentials, effective masses...). So the absorption signal accounts for both

transitions. However, it has been established through calculations that the strain preserves

the oscillator strength of the fundamental transition (Γ7CH-CB). Therefore it can be con-

cluded that our MQW design with the use of relaxed buffer is promising for the fabrication

of surface-emitting LED or lasers in DUV region.

The results of this section can be found in [150].

4.3 10- and 20-period MQWs

The results shown in the last section is an example for 4-period MQWs. In this section, 10-

and 20-period MQWs would be analyzed and compared in the aspect of structural quality

and emission properties. These samples would be used for the processing of devices by

depositing high-reflective dielectric DBR on the top and the bottom after thinning down

sapphire.

The XRD symmetric 2θ-ω scans and unsymmetric RSMs can be seen in Fig. 44. Satel-

lite peaks associated with MQWs interfaces can be clearly observed for both. The MQWs

are pseudomorphically strained on the AlGaN relaxed buffer based on (1 1 -2 4) RSMs.

82

Figure 44: Symmetric 2θ-ω scans and (1 1 -2 4) RSMs of (a) 10-period and (b) 20-periodMQWs grown on a relaxed AlGaN buffer on AlN templates.

According to the simulation, the buffer containing 0.58 (± 0.01) Al has 70% relaxation,

the wells of both samples contain 0.38 (± 0.015) Al with thickness of 1.3 nm (± 0.2 nm),

and the barriers contain 0.58 (± 0.015) Al with thickness of 9.3 nm (± 0.2 nm). Figure 45

shows that the 20-period MQWs exhibit obvious growth terraces beyond the V-pits. The in-

fluences of V-pits would be discussed in the section 4.4. Cross-sectional HAADF-STEM

image of 10-period MQWs and corresponding chemical mapping were demonstrated in

Fig. 46. Based on the chemical mapping, the ultra-thin wells contain 0.37 (± 0.016) Al and

thickness was around 1.5∼1.7 nm, which confirmed the XRD simulations.

The optical properties were investigated by CL at room temperature with different ex-

citation energy. The wavelength of 20-period MQWs is 17 nm red-shifted, which can be

due to the thickness variation of ultra-thin wells between samples. Based on our studies

on different MQW samples, a small change of well thickness (e.g. from 1.1 nm to 1.5

83

Figure 45: AFM images of 20-period MQWs grown on a relaxed AlGaN buffer on AlNtemplates.

nm) can cause an obvious shift of the emission wavelength. However, the determination

of well thickness can not be accurate enough to identify this small variation. Along the

increase of excitation energy, the wavelengths stay almost constant. The intensity increases

first and then decreases along with the increase of penetration depth. 10-period MQWs

achieve maximum emission intensity under 4 keV with penetration depth of around 40 nm

and 20-period MQWs achieve maximum under 6 keV with penetration depth of around 70

nm. The FWHM demonstrates an inverse “S” shape, which might be related to the carrier

carrier dynamics and also penetration depth in the MQWs.

SEM image and corresponding hyperspectral CL mapping of 20-period MQWs are

shown in Fig. 48. It scans over the large area ( 228 × 171 µm2), and the variation of

emission wavelength is only 0.9%, the variation of FWHM is 10%, and the variation of

intensity is 8%, which indicate the good uniformity of the MQW growth.

4.4 Defects in MQWs and their influence on DUV emission

Compared with InGaN-based visible laser diodes or light emitting diodes (LEDs), the ef-

ficiency of DUV light sources based on AlGaN material system is much lower. One of

84

Figure 46: (a) Cross-sectional HAADF-STEM image of 10-period MQWs grown on arelaxed AlGaN buffer; (b) Compositional mapping obtained from the STEM image; (c) Alcontent distribution along the profiles marked in (a).

85

Figure 47: Comparison of (a) CL emission wavelength, (b) FWHM and (c) emission in-tensity at room temperature between 10- and 20-period MQWs grown on a relaxed AlGaNbuffer on AlN templates.

86

Figure 48: CL hyperspectral mapping of 20-period MQWs at room temperature.

the biggest issues is the rapid degradation of structural quality for Al(Ga)N epitaxial lay-

ers with increased Al content. Bulk AlN substrates could be good candidates to achieve

low-threshold optically pumped DUV lasers due to small lattice mismatch with epitaxial

layers and low dislocation density below 5×108 cm−2 [21, 43, 44], but they suffer from high

impurity absorption, high cost and limited availability. LED and laser devices were also

obtained on AlN templates grown on sapphire [46, 47], but threading dislocation density

is still above 109 cm−2, limiting performance. Besides threading dislocations (TDs), the

impurity (such as oxygen) effect is greater in AlN epitaxial layers than in GaN due to low

diffusion length of Al atoms (high sticking coefficient) and increased affinity of Al to ox-

idize, which cause rather high density of defects in AlGaN layers with Al content larger

than 50% [22, 23]. Unlike the defects in InGaN/GaN MQWs which have been widely

studied (even though the influence of defects on emission efficiency is still controversial)

[151–156], only a few reports have focused on the study of the typical defects in metal

organic vapor phase epitaxy (MOVPE) grown AlGaN MQWs on AlN templates for DUV

87

Figure 49: 2θ-ω scans of (a) sample #1 and (b) sample #2.

devices and especially their influence on the optical emission [157–160].

In this section, the threading dislocations and V-shape pits in AlGaN MQW structure

grown on AlGaN buffer on AlN templates were characterized by different methods that

allow a discussion of the potential origin of dislocations and V-pits [161]. MQWs samples

with different defect densities were prepared in order to study the optical influence of the

defects on DUV emission at 280 nm.

4.4.1 Structural investigations of defects

AlN templates grown on c-axis sapphire with different threading dislocation densities were

used as substrates for the same structure, one with high dislocation density (template #1)

88

and the other with low dislocation density (template #2). The samples grown on the same

type of template in different runs have comparable defect density, which confirmed repro-

ducibility. The samples grown on different types of templates have significant difference

in V-pit density and they were labeled as sample #1 (on template #1) and sample #2 (on

template #2) for this study. Figure 49 shows (0 0 0 2) 2θ-ω scans of both samples. Satellite

diffraction peaks (SL) associated with the quantum wells can be observed which indicates

abrupt interfaces between wells and barriers. Based on (1 1 -2 4) reciprocal space mapping,

the buffer containing 0.58 (± 0.01) Al has 70% relaxation. According to the simulation,

the wells of sample #1 contain 0.38 (± 0.015) Al with thickness of 1.5 nm (± 0.2 nm),

while the barriers contain 0.58 (± 0.015) Al with thickness of 10.3 nm (± 0.2 nm). For the

sample #2, the wells contain 0.38 (± 0.015) Al with thickness of 1.3 nm (± 0.2 nm), while

the barriers contain 0.58 (± 0.015) Al with thickness of 9.3 nm (± 0.2 nm).

The FWHM angles measured for AlN templates and thick AlGaN layers (buffer +

MQWs) of two samples are shown in Fig. 50. The screw and edge TD densities calcu-

lated by the method above are summarized in table 1. For both samples, extra TDs were

not generated within the relaxed AlGaN buffer layers. The AlN template used for sample

#1 exhibits 3 times higher screw and edge TD densities than sample #2, and the density

of pits on the surface is around 30 times higher as seen in SEM images. These holes have

hexagonal shape, and the ultra-high density of the holes in sample #1 leads to the surface

inhomogeneities. On the contrary, sample #2 with much lower pit density has a flat sur-

face (apart from holes) with root-mean square roughness (RMS) as small as 0.5 nm by

AFM measurements. The slight difference of thickness may be due to unintentional drift

between the runs.

In order to study the origin of these holes and their influence on the MQWs growth, the

cross-section HAADF-STEM was performed and images of sample #1 are shown in Fig.

51. In the flat area of Fig. 51(a), the 10-period MQWs have good uniformity and periodic-

ity. The thickness of barriers and wells agrees with XRD simulation results. In Fig. 51(b),

89

Figure 50: (a) Skew symmetric ω scans and (b) SEM images of sample #1; (c) skew sym-metric ω scans and (d) SEM images of sample #2.

Table 8: Estimated defect densities (FWHM determination and linear fitting lead to anestimated error of 15% for threading dislocations) for sample #1 and sample #2.

Screw TDs (cm−2) Edge TDs (cm−2) V-pits (cm−2)

Sample #1AlN template ∼ 7.3 × 109 ∼ 1.7 × 1011

∼ 2 × 109AlGaN buffer ∼ 6.7 × 109 ∼ 1.5 × 1011

Sample #2AlN template ∼ 2.0 × 109 ∼ 4.9 × 1010

∼ 7 × 107AlGaN buffer ∼ 2.1 × 109 ∼ 3.5 × 1010

90

it is clearly observed that the threading dislocations originate from the nucleation layer

at the interface of AlN and sapphire. After certain thickness, the dislocations are mostly

aligned along the [0 0 0 1] axis. These dislocations can form low angle grain boundaries

and lead to a twist of the lattice. The V-pits near the surface are always associated to these

grain boundaries. Figure 51(c) is the plan-view HRTEM images of the AlN template in

which some grains with size from 50 to several hundred nanometers can be identified. The

atomic resolution STEM image with higher magnification is shown in Fig. 51(d) to get

rid of the Moire fringes. These grains are separated by low-angle boundaries with small

misorientation of 1 ∼ 2 degrees. During regrowth of AlGaN layers and MQWs, the thread-

ing dislocations and grain boundaries propagate into the following layers and penetrate the

whole structure.

By looking into the V-shape pits in Fig. 51(e) with higher magnification, the apex of the

holes is connected to extended defects which are c-axis oriented and inherited from the AlN

layer. These lines should be edge threading dislocations or mixed threading dislocations

since the image is taken along the <1 1 -2 0> zone axis. The possible cause of V-pits could

be that impurities tend to segregate into dislocations, which may locally impede growth

resulting in a small indentation. Once the facets of slow growth planes are formed due to

these indentations, a V-pit can be generated [162, 163]. Compared with the sample #2, the

sample #1 has higher edge threading dislocation density which increases the probability for

forming pits.

As shown in Fig. 51(e) and 51(f), the facets of the holes are inclined about 62◦ to the

c plane, which corresponds to {1 0 -1 1} facets. The growth rate on these planes is much

slower than on the c plane [162, 163], so a V-shape hollow hole is formed. Figure 51(f)

is an example of 4-period MQWs for investigating the sidewall growth in a V-pit. On the

V-pit sidewall which is near c-plane surface, the weak contrast of QWs can be observed.

Both the thickness of wells and barriers here are almost one third of the ones grown on the c

plane. This small area of QW growth on the sidewall may lead to some additional emission

91

wavelengths with weak intensity giving rise to main peak broadening. In the area closer

to the center of the hole, the wells are merged into the barriers and cannot be identified.

The V-pits interrupt the superlattice growth and may lead to non-uniformity of 2D MQW

properties.

4.4.2 Optical influence of defects in AlGaN MQWs

The optical properties of the MQWs samples were investigated. The PL measurements

were done from 5 K to 275 K with a cw laser (266 nm) and a weak excitation power of

1W/cm2 in order to maintain the thermal equilibrium condition and to avoid saturation of

emission intensity. The spot diameter of PL is around 100 µm. For transmission measure-

ments, the size of probed area is around 0.02 mm2. Figure 52(a) shows the PL at 80 K

and transmission spectrum at 77 K. Sample #1 shows a well-defined emission at 282 nm.

The absorption edge of the wells is at 281 nm and absorption edge of barriers is at 260 nm,

which is in a good agreement with the design and with the compositions determined by

XRD. For the sample #2, the PL emission is at 278 nm while absorption edge is 275 nm

for the wells and 260 nm for the barriers. The absorption edge of sample #2 is 6 nm blue-

shifted which can be due to the thinner wells according to XRD simulation. The emission

of sample #1 has full width at half maximum (FWHM) of 12.5 nm, while sample #2 has

FWHM of 7.2 nm indicating a lower inhomogeneous broadening, which is consistent with

transmission measurements. The broadening of emission peak in sample #1 can be caused

by the local variations of thickness or composition in MQWs which occurred in the high

density of V-pit sidewalls near the c-plane surface, as shown in Fig. 51(f).

Figure 53 shows the PL integrated intensity as function of the inverse of temperature as

well as Arrhenius fitting curves. The fitting equation is as follows [164]:

IPL(T ) =I0

1 + Aexp(−EA

kT) + Bexp(−

EB

kT), (39)

where I is the integrated intensity with arbitrary unit, and I0 is a fitting parameter which can

be seen as the integrated intensity when T is close to 0 K. A and B are fitting constants for

92

Figure 51: (a) Cross-section HAADF-STEM image of sample #1 showing 10-periodMQWs without dislocations; (b) low-magnification bright field (BF) STEM image show-ing origin of defects; (c) plan-view HRTEM image of AlN template on sapphire showinggrains; (d) high magnification image of grain boundaries; (e) high-magnification BF imageon V-shape pits; (f) high-magnification HAADF-STEM image of 4-period MQWs showingsidewall of V-pits.

93

Figure 52: Transmission spectra at 77 K and macro-PL at 80 K for (a) sample #1 and (b)sample #2.

94

Figure 53: PL integrated intensity as function of temperature (with a cw laser and a weakexcitation of ∼ 1W/cm2) and Arrhenius fitting for (a) sample #1 and (b) sample #2; (c) isthe curve of IQE versus temperature based on Arrhenius equation fitting.

two nonradiative recombination channels: the first mechanism (channel A) is for detrapping

of localized excitons which dominates at low temperature range, and the second (channel B)

is for escaping out of the wells which dominates at high temperature range. EA and EB are

corresponding activation energies required for two paths of carriers escape from localized

states into the non-radiative recombination centers. The internal quantum efficiency (IQE)

can be compared by defining IQE=I(T)/I0 [165, 166]. In this case, the efficiency is assumed

to be 100% at 5 K, which is reasonable since the intensity was almost constant (to be I0) in

the temperature range of 5 to 50 K indicating almost no influence of nonradiative process

in this range.

Based on Arrhenius equation fitting, sample #1 has 29 meV for EA and 96 meV for

95

EB. Sample #2 has 49 meV for EA and 150 meV for EB. There is a larger energy barrier

of nonradiative processes for sample #2. Figure 53(c) shows the IQE variation along with

the temperature for two samples based on Arrhenius equation. At 77 K, the IQE of sam-

ple #1 decreased to 80% while sample #2 still has IQE of 94%. This can be explained by

the increased concentration of nonradiative centers caused by higher dislocation density

in sample #1 [167, 168]. In the high temperature range, the IQE of both samples drops

dramatically due to the high level of dislocation densities. Near room temperature, the

error bar becomes large because emission intensity is weak and the residual laser contri-

bution to the luminescence is not negligible. Both samples have IQE of 0.5 ∼ 1% near

room temperature, which is in agreement with simulated value for this level of dislocation

densities [169]. To increase the IQE at room temperature, the quality of AlN templates

should be further improved [118, 119] to reduce threading dislocation densities. The 30

times higher density of V-defects in sample #1 doesn’t seem to affect the PL behaviour. It

could be explained by the fact that the QWs are almost absent in the V-pits according to

Fig. 51(f), in other words, the contribution of this part to the intensity (I) of main emission

peak is negligible. But it should be noted that these pits may lead to current leakage as

what has been observed in InGaN MQWs [153] and to lower external emission efficiency

of current-injected LEDs or lasers [158].

The CL spectra shown in Fig. 54 were collected by scanning over an area ∼ 8 µm ×

11 µm at room temperature, giving the average information of the sample (including flat

areas and rough areas with pits). The measurements were done under excitation from 3

keV to 7 keV. Figures 54(a) and 54(b) show the spectra under 7 keV which corresponds

to a penetration depth of the electron beam of typically 90 nm. Emission from wells is

located at 286 nm with FWHM of 16 nm for sample #1 and at 276 nm with FWHM of

13 nm for sample #2. Besides the emission from MQWs, both samples show weak defect

band emission at 333 nm. By comparing the emission intensity from the wells and the

defect band, the ratio of IMQWs/Ide f ects is 4.5 for sample #1 and 24 for sample #2 which

96

Figure 54: CL spectra under 7 keV at room temperature for (a) sample #1 and (b) sample#2; (c) is FWHM of QW emission peaks under different electron beam energies at roomtemperature.

may be attributed to lower dislocation density and the existence of more flat surface area

with good quality QWs which can interact with the electron beam. The comparison of

FWHM between the two samples in Fig. 54 confirms again the higher emission uniformity

of sample #2 with the fewer V-pits.

97

CHAPTER 5

DISTRIBUTED BRAGG REFLECTOR: SIMULATIONS ANDREALIZATION

Chapter 5 focuses on the theoretical simulation and the experimental realization of DUV

DBRs that apply BAlN/Al(Ga)N heterostructures. The simulation of the ideal DBR struc-

tures is based on the transfer-matrix method presented in the first section. The reflector

software was developed by Dr. Abid in his thesis in 2013 [9]. The design of DBRs is

discussed in the second section.

In the third section, different factors leading to the degradation of DBR performance

including surface roughness, interface roughness and strain are introduced into the reflector

software, and their influences on the DBR reflection are investigated theoretically. Then

the experimental results of BAlN/Al(Ga)N DBRs obtained are presented in the last section.

The influences of the structures and growth conditions are studied. Based on the detailed

characterizations, the simulations with the input of quality parameters are performed and

then compared with the experimental spectra .

5.1 Transfer-matrix simulations of DBRs

This section describes the principles and methods for simulating the specular reflection on

the surface of distributed Bragg mirrors.

The tangential components of the electric and magnetic fields are continuous across the

interface, so it should satisfy the boundary conditions as follows:Ei1 + Er1 = Et1 + E′i1

Bi1/µ1 − Br1/µ1 = Bt1/µ2 − B′i1/µ2 ,

(40)

where E and B are electric field and magnetic field, respectively, and µ is the permeability.

It is known that:

98

Figure 55: Schematic of electromagnetic wave propagation in the DBR structure.

B =√µεE = n

√µ0ε0E = γE , (41)

where ε is the permittivity, n is the refractive index and γ is the inverse of the wave velocity

in the film. For non-magnetic media, µ is equal to µ0. So the Eq. 40 can be written as:

Ei1 + Er1 = Et1 + E′i1

γ1(Ei1 − Er1) = γ2(Et1 − E′i1) .(42)

Considering the wave propagation in the layer 2, we have:E′i1 = Er2e− jϕ2

Et1 = Ei2e jϕ2 ,

(43)

where the phase difference in the film ϕ=2πnt/λ0. So the Eq. 42 can be converted to:Ei1 + Er1 = Ei2e jϕ2 + Er2e− jϕ2

γ1(Ei1 − Er1) = γ2(Ei2e jϕ2 − Er2e− jϕ2) .(44)

Then Ei1 and Er1 at the 1st interface can be expressed as the function of the Ei2 and Er2 at

99

the 2nd interface: Ei1

Er1

=

γ1 + γ2

2γ1e jϕ2

γ1 − γ2

2γ1e− jϕ2

γ1 − γ2

2γ1e jϕ2

γ1 + γ2

2γ1e− jϕ2

Ei2

Er2

=

12

1 +

γ2

γ11 −

γ2

γ1

1 −γ2

γ11 +

γ2

γ1

e jϕ2 0

0 e− jϕ2

Ei2

Er1

.(45)

Then Wi,i+1 can be defined as the refraction term representing the interface between the ith

and (i+1)th layer which only relates to the refractive indices of two layers , and Ui is the

phase matrix representing the ith layer the wave is propagating through, which relates to the

refractive index and optical thickness of the layer.

Wi,i+1 =12

1 +

γi+1

γi1 −

γi+1

γi

1 −γi+1

γi1 +

γi+1

γi

=12

1 +

ni+1

ni1 −

ni+1

ni

1 −ni+1

ni1 +

ni+1

ni

Ui =

e jϕi 0

0 e− jϕi

, ϕi = 2πniti/λ0 ,

(46)

Considering the incidence term from the substrate to the bottom layer Eis=0, the surface

incidence and reflection terms are derived to be: Ei0

Er0

= W0,1U1W1,2U2...Wk−1,kUkWk,k+1

Eis

Ers

=

s11 s12

s21 s22

0

Ers

. (47)

The reflection R is equal to:

R =

∣∣∣∣∣Er0

Ei0

∣∣∣∣∣2 =

∣∣∣∣∣ s22

s12

∣∣∣∣∣2 . (48)

By giving the refractive indices and thicknesses of each layers in DBRs, the specular re-

flection on the top surface can be obtained by calculating the elements of transfer matrix.

The optical thickness of each layer is normally one-quarter of the wavelength so that one

period is equal to λ/2 to have constructive interference for the reflection on the surface.

100

Figure 56: Experimental refractive index of BAlN versus incident wavelength [8, 9].

5.2 Design of BAlN/AlGaN DBRs

The Reflector software has been programmed by Dr. Abid based on the transfer matrix

method described in the last section to simulate the reflection of III-nitride epitaxial DBRs

in the ideal case [9]. In the software, the refractive indices of AlGaN are derived from the

equations reported by Brunner [10]:

n(AlxGa1−xN) =

√C(x) +

A(x)(2 −√

1 + y −√

1 − y)y2

C(x) = −(2.2 ± 0.2)x + (2.66 ± 0.12)

A(x) = (3.17 ± 0.39)√

x + (9.98 ± 0.27)

y =hν

Egx.

(49)

And the refractive indices of BAlN used in the simulation are experimental values taken

from Ref. [9, 64], as shown in Fig. 56.

In this project, BAlN/AlGaN structure is proposed for DBRs reflecting at DUV wave-

lengths. Compared with the conventional AlGaN/AlGaN structure, it has larger refractive

index contrast so that the reflectivity and stopband width can be increased. Additionally,

101

the tensile strain of BAlN layers can compensate the compressive strain induced in AlGaN

layers.

Several factors need to be considered during the design of DBRs. Firstly, large refrac-

tive index contrast in the two layers is required, which means more boron in BAlN and

more Ga in AlGaN. But this will lead to larger lattice mismatch with the AlN template.

Meanwhile, more Ga in AlGaN would lead to more absorption of DUV wavelengths. The

number of the periods or the total thickness also needs to be paid attention to. Too thick

structure is not favorable for the quality concern, so it would be limited below 20 periods.

The last point is that high boron incorporation would be difficult for MOVPE techniques.

In order to avoid absorption by AlGaN band tail states, the composition of AlGaN

should be chosen in the way that its absorption coefficient for the targeted wavelength

should be below 103 cm−1, or the band-edge wavelength is at least 40 nm shorter than

the targeted wavelength. The AlGaN absorption coefficient as function of the energy for

different Al content is shown in Fig. 57. Based on this criterion, the AlGaN layers should

contain more than 70% Al.

Choosing AlGaN layer with 70% and 80%Al, the simulations were done for the struc-

tures with different boron compositions in BAlN layers, and with different number of pe-

riods of stacks.The results are shown in Fig. 58. It can be seen that in the range of 0 to

15% boron, the reflectivity increases up to 4% boron. After that, the reflectivity keeps con-

stant. Considering all the factors mentioned before, 4% boron was proposed. Also, by this

composition, the lattice mismatch between BAlN and AlN is -0.74%, which can almost

compensate the compressive strain in AlGaN (around 0.64%). So the proper structure is:

33 nm BAlN containing 4% boron and 28 nm AlGaN containing 70% Al.

The simulation and the design are based on the ideal case and there would be deviations

in the real experiments, especially considering the growth challenges and the fact that the

quality of the epitaxial layers can significantly affect the DBR performance. In fact, the

DBRs with different compositions were performed in the experiments for the studies.

102

Figure 57: AlGaN absorption coefficient as function of the energy for different Al content[10].

5.3 Simulation of structural quality factors

The simulation in the last two sections is for an idea case and it means that the DBR

structure has perfect quality without any roughness, defects or any strain in the layers.

However, in the real case, thick DBR structures usually suffer from quality issues. In this

section, different factors involving the structural quality and strain state have been input into

the simulation and software, in order to understand their influences on the performance of

DBRs. The factors considered include roughness (surface and interface roughness) and

lattice strain. The methods to introduce those factors are presented in this section and their

impacts on the reflecting behavior of DBRs are discussed.

5.3.1 Roughness

The roughness including surface roughness and interface roughness cause light scatter-

ing on the surface and interface of the DBRs, which reduces the specular reflectance. To

consider the scattering effect at each interface, the large-scale roughness model should be

103

Figure 58: Simulated reflectivity of (a) BxAl1−xN/Al0.70Ga0.30N structure and (b)BxAl1−xN/Al0.80Ga0.20N structure.

104

Figure 59: Lattice mismatch of AlGaN/AlN and BAlN/AlN.

applied [170]. The effect of the corresponding roughness harmonic on the amplitude re-

flection and transmission in Fig. 55 can be described as [170]:r(w.roughness) = r(w.o.roughness)(1 − 2k2n2

1σ2)

t(w.roughness) = t(w.o.roughness)[1 − 0.5k2(n1 − n2)2σ2] ,(50)

where r =Er1

Ei1, t =

Et1

Ei1, k =

2πλ

and σ is the root mean square roughness (RMS). If it is

assumed that rrmsi = (1 − 2k2n2iσ

2) and trmsi = 1 − 0.5k2(ni − ni+1)2σ2, the interface term

Wi,i+1 in Eq. 48 can be replaced by:

Wi,i+1 =12

(1 +

ni+1

ni)

1trms

(1 −ni+1

ni)rrmstrms

(1 −ni+1

ni)rrmstrms

(1 +ni+1

ni)

1trms

. (51)

Figure 60 shows the simulation results for 20-period B0.04Al0.96N (33 nm) /Al0.70Ga0.30N

(28 nm) DBR after inputting roughness values (the surface roughness is considered to be

equal to the interface roughness). The roughness can lead to both reflectivity reduction and

phase shift of the reflectance spectrum. The inset of Fig. 60 gives the relationship between

the central wavelength reflectivity and the roughness. When the roughness increases from

105

Figure 60: Simulated reflectance of 20-period B0.04Al0.96N (33 nm) /Al0.70Ga0.30N (28 nm)DBR with different values of roughness; Inset shows the curve of the central wavelengthreflectivity versus the roughness.

0 to 6 nm, there is only several percent decrease of the reflectivity. However, when the

roughness exceeds 8 nm, the reflectivity of this structure drops dramatically.

5.3.2 Influence of lattice strain

In order to get a good quality of DBR, the structure is expected to be coherently-strained

on the substrate, since the relaxation would lead to dislocations or cracks. In this case,

each layer is under compressive or tensile strain. In this subsection, the relation between

in-plane strain of the DBR layers and the reflectivity is investigated and discussed.

The first question would be how to input the strain factor in the simulation. The general

idea is that: the lattice strain influences the energy gap Eg, and then changes the refractive

index of the layer n. It is assumed that the variance of n due to lattice strain ε is relatively

small compared with the original value. Thus, the refractive index variance ∆n of ternary

material AxB1−xN can have a linear relationship with the variances of its binary compounds:

∆n = x · ∆n(AN) + (1 − x) · ∆n(BN) . (52)

106

Considering the correlation between refractive index, energy gap and lattice strain, this

equation can be changed into:

∆n = x ·∂n(AN)∂Eg

∆Eg(AN) + (1 − x) ·∂n(BN)∂Eg

∆Eg(BN)

= [x ·∂n(AN)∂Eg

∂Eg(AN)∂ε

+ (1 − x) ·∂n(BN)∂Eg

∂Eg(BN)∂ε

]∆ε .(53)

The DBR is a multi-layered structure in which the strain state in each layer is difficult

to determine when the structure is relaxed. So the structure is assumed to be pseudomor-

phically strained. The in-plane lattice of the structure is either equal to the in-plane lattice

of the substrate, or it can be obtained by asymmetric RSMs.

Then it needs to find out the equation of parameter “n” in the form of Eg and parameter

Eg in the form of strain. As the linear interpolation is taken into account, only equations of

binary alloys instead of complex ternaries are required.

A. Hafaiedh [171] has noted that the empirical expression proposed by Herve and Van-

damme [172] gives better agreement with known data obtained for refractive indices in the

case of III-V semiconductors. And this model is used here:

n =

√1 + (

13.6Eg + 3.4

)2

∂n∂Eg

= −13.62

(Eg + 3.4)2√

(Eg + 3.4)2 + 13.62.

(54)

According to S. Krishnanukutty [173], the change in the energy gap as the function of

the strain in a direction can be generally written as:

∆Eg = α · ε . (55)

In a hexagonally symmetric crystal:

α =2a(C33 −C13)

C33= 2a ·

1 − 2ν1 − ν

, (56)

where a is the hydrostatic deformation potential given by:

a = −K ·dEg

dp. (57)

107

Table 9: Parameters used of III-nitrides for Eq. 59BN AlN GaN

Eg [eV] 5.9 6.28 3.42ν 0.19 [174] 0.203 [106] 0.183 [106]

E [GPa] 400 [174] 330 [175] 290 [106]dEg/dε [eV/GPa] 0.006 [176] 0.049 [177] 0.047 [178, 179]

K is the bulk modulus which can be converted into an expression of Young’s modulus E

and Poisson ratio ν:

K =E

3(1 − 2ν). (58)

So, the derivative of Eg with respect to the strain can be written as:

dEg

dε= −

2E3(1 − 2ν)

·1 − 2ν1 − ν

·dEg

dp= −

2E3(1 − ν)

dEg

dp. (59)

For AlN, GaN, BN, the parameters from the literatures are summarized here in Tab. 9.

In this way, the derivative of refractive indices of III-nitrides with respect to the strain

can be obtained:

∂n(BN)∂Eg(BN)

∂Eg(BN)∂ε

= −13.62

(Eg + 3.4)2√

(Eg + 3.4)2 + 13.62·−2E

3(1 − ν)dEg

dp

= −13.62

(5.9 + 3.4)2√

(5.9 + 3.4)2 + 13.62·−2 · 400

3(1 − 0.19)· 0.006

= 2.91

∂n(AlN)∂Eg(AlN)

∂Eg(AlN)∂ε

= −13.62

(Eg + 3.4)2√

(Eg + 3.4)2 + 13.62·−2E

3(1 − ν)dEg

dp

= −13.62

(6.28 + 3.4)2√

(6.28 + 3.4)2 + 13.62·−2 · 330

3(1 − 0.203)· 0.049

= 1.60

∂n(GaN)∂Eg(GaN)

∂Eg(GaN)∂ε

= −13.62

(Eg + 3.4)2√

(Eg + 3.4)2 + 13.62·−2E

3(1 − ν)dEg

dp

= −13.62

(3.42 + 3.4)2√

(3.42 + 3.4)2 + 13.62·−2 · 290

3(1 − 0.183)· 0.047

= 0.26 .(60)

108

For the ternary alloys, the expressions are as follows:

AlxGa1−xN : ∆n = [1.60x + 2.91(1 − x)] · ∆ε

BxAl1−xN : ∆n = [0.26x + 1.60(1 − x)] · ∆ε

BxGa1−xN : ∆n = [0.26x + 2.91(1 − x)] · ∆ε .

(61)

Figure 61(a) shows the theoretical reflectivity of 20-period AlN/AlxGa1−xN DBR with

different Al content (x) in the AlGaN layer with or without consideration of the strain effect

in the simulation. Fig. 61(b) is the same case for 20-period BxAl1−xN/Al0.7Ga0.3N DBR. It

means that the input of the strain effect only leads to a small reduction of reflectivity which

is less than 1%. The strain in the layers doesn’t affect the reflectivity in an obvious way by

influencing the refractive indices. But it should be paid attention to that large strain in the

layers may degrade the structural quality and therefore the device performance.

The interface of the reflector software after inputting surface roughness, interface rough-

ness and strain is shown in Fig. 62.

5.4 Realization of BAlN/Al(Ga)N DBRs for DUV

This section describes the realization of BAlN/Al(Ga)N DBRs in T-shape MOVPE reac-

tor. Related characterizations were performed and analyzed. The influences of increasing

TEB/III ratio for BAlN layers on structural and optical characteristics of DBRs as well as

the influences of replacing AlN layer with AlGaN were studied. The experimental results

are also compared with the simulations by using parameters determined from characteriza-

tions.

5.4.1 Growth conditions for DBRs

The section 3.2 analyzed structural features of BAlN layers grown at 1000 ◦C and at low

temperature of 650 ◦C with annealing at 1020◦C by FME method. Figure 63 compares

the cross-sectional HAADF (BF) images of the BAlN layers or heterostructures grown in

those conditions with the heterostructures grown at 1000 ◦C in a continuous way. By FME

109

Figure 61: Simulated reflectivity spectra with and without consideration of the strain for(a) 20-period AlN/AlxGa1−xN DBR and (b) BxAl1−xN/Al0.70Ga0.30N DBR.

110

Figure 62: Reflector software interface with factors of surface roughness, interface rough-ness and strain.

111

method, crystalline quality can be improved since the grain size is increased and columns

are formed. But in this case, the interface of BAlN and AlN layers would be destroyed,

which can be a disaster for DBR. For example, the structure of 5-period BAlN (32 nm) /

AlN (25 nm) didn’t have any reflection peak at around 260 nm as it is supposed to.

Meanwhile, by looking into the structure grown at 1000 ◦C simply in a continuous

way (Fig. 63(c)), the BAlN layer is nanocrystalline and the small-size crystallites maintain

an accepted interface as well as a clear contrast between two materials. In this work, the

DBRs for DUV region have been realized under this condition. More details and analyses

of STEM images about the DBR structure are presented in subsection 5.4.3.

5.4.2 BAlN/Al(Ga)N DBRs with reflection at DUV wavelengths

Firstly, a series of 18-pair BAlN/Al0.72∼0.76Ga0.28∼0.24N DBRs have been grown. The aver-

age growth rate for the BAlN/AlGaN structure is around 2 µm/h. Different TEB/III ratios

were used for BAlN layers in order to study the influence of boron incorporation on the

performance of DBRs.

It is known that the surface of AlN or AlGaN with high Al content (>70% Al for DUV

design) would have high density of V-pits due to the low diffusion length of Al atoms

and high affinity to oxygen at 1000 ◦C, as shown in Fig. 64. That’s to say, for simple

AlGaN/AlN structures, the flat areas apart from pits have 2D morphology, but the surface

would be isolated into small islands by the high density pits as shown in Fig. 64(c). The

heterostructure growth would occur on the sidewalls of the pits and the island morphology

would scatter the light on the surface. For these AlGaN (with more than 70% Al) /AlN

DBRs designed for DUV wavelengths, no reflection was obtained.

Figure 65 shows the SEM images, XRD experimental spectra with fittings and reflec-

tion spectra of the BAlN/AlGaN DBRs using three different TEB/III ratios. The composi-

tion of boron used in the XRD fittings is estimated by assuming that it has a simple linear

relationship with TEB/III ratio, and it is 15% for TEB/III=39% according to EDX measure-

ments. When TEB/III=7%, very little boron was introduced into the layer (might be around

112

Figure 63: Cross sectional HAADF-STEM (BF) images of (a) 75 nm BAlN grown at 650◦C with annealing at 1000 ◦C by FME method, (b) 5-pair BAlN (32 nm) / AlN (25) nmgrown at 1000 ◦C by FME and (c) 18-pair BAlN (32 nm) / AlGaN (24 nm) grown at 1000◦C in a continuous way.

113

Figure 64: SEM images of (a) 30 nm Al0.82Ga0.18N and (b) 5-period AlN (30 nm) /

Al0.82Ga0.18N (30 nm) grown on AlN/sapphire templates; (c) cross-sectional SEM imageof the sample in (b).

114

2∼3%). The morphology of the structure is in the transition stage between 2D morphology

isolated by pits and 3D nanocrystalline morphology. For this morphology, no reflection

was obtained as what was observed for AlN/AlGaN structures. But the XRD spectrum

exhibit good fringes relating to periodicity of heterostructures since it’s still maintained

monocrystalline.

When the TEB/III ratio was increased, the structure becomes polycrystalline and crys-

tallites spread on the whole surface. In the XRD, only the peak relating to the first monocrys-

talline AlGaN layer before BAlN growth can be observed, except that there is a small peak

at 37.7 ◦ in the XRD of TEB/III=26% which can be related to (1 -1 0 1) facet of AlGaN.

For TEB/III=26%, 30% reflection was obtained for the central wavelength of 267 nm

with a stopband-width of 14 nm. For TEB/III=39%, 55% reflection was obtained for the

central wavelength of 279 nm with a stopband-width of 20 nm. The increase of the re-

flection and stopband-width is due to higher refractive index contrast introduced by more

boron incorporation and less absorption at longer wavelength.

DBRs were also grown with a lower growth rate of 500 nm/h, since the low growth

rate can increase the mobility of Al and B elements, and hence improve the homogeneity

and interface abruptness [180]. The RMS is around 9 nm over 20 × 20 µm2 while the

RMS of similar structures grown at 2 µm/h is around 16 nm. The experimental reflec-

tion spectra along with the simulations of three DBR structures are summarized in Fig.

66. The simulations here only considered ideal case without roughness or strain, and the

comparison between experimental data and simulations considering quality factors would

be demonstrated in the next subsection. The composition and thickness of AlGaN layers

were determined by XRD. The composition and thickness of BAlN were determined by

EDX (next subsection) and approximate calculation from reflection peak wavelength (the

optical thickness of each layer should be one quarter of the central reflection wavelength).

For BAlN/AlN structures, the experimental results are very close to the simulations. From

central wavelength of 260 nm to 282 nm, the reflection was increased from 48% to 70%.

115

Figure 65: 18-pair BAlN/Al0.72∼0.76Ga0.28∼0.24N DBRs with different TEB/III ratios (Thecomposition of boron used in the XRD fittings is estimated by an assumption that it has alinear relationship with TEB/III ratio, and it is 15% for TEB/III=39% according to EDXmeasurements).

116

Figure 66: (a) 18-pair BAlN (29 nm) / AlN (29 nm) DBRs reflecting at 260 nm; (b) 18-pair BAlN (33 nm) / AlN (32 nm) DBRs reflecting at 280 nm; (c) 18-pair BAlN (33 nm) /

Al0.8Ga0.2N (24 nm) DBRs reflecting at 265 nm.

At central wavelength of 260 nm and 265 nm, by replacing AlN with Al0.8Ga0.2N, the re-

flection was also increased from 48% to 70% and the stopband width was increased from

12 nm to 19 nm (more than 50%) which confirmed better performance of BAlN/AlGaN

structure than BAlN/AlN structure. In order to have a simulation more close to the real

case, additional characterizations are required to determine the thickness for each layer,

boron content, roughness values and so on.

5.4.3 Characterizations of DBRs and reflectance comparison with simulations

Taking the sample in Fig. 66(c) as an example, more structural analyses are presented in

this subsection. Experimental reflection spectrum is compared with the simulation in ideal

117

Figure 67: (a) HAADF-STEM (BF) image of 18-pair BAlN/Al0.8Ga0.2N DBRs reflectingat 265 nm. (b), (c) and (d) are TEM images with different magnifications.

case and with the simulation considering quality factors. The parameters for simulations

were obtained from characterizations.

Cross-section STEM images are shown in Fig. 67. It’s clear that BAlN layers consist

of nanocrystals with size of 2∼3 nm instead of forming bigger columns as the ones grown

by FME method. The following AlGaN layers inherit this nanocrystalline feature but tends

to form bigger grains (5∼10 nm) since Ga atoms have larger diffusion length than B atoms.

These grains increase the interface roughness (Fig. 67(b)). The thickness is estimated to be

118

Figure 68: Cross-sectional EDX mapping of 18-pair BAlN/AlGaN DBR reflecting at 265nm.

31 nm for BAlN and 27 nm for AlGaN.

EDX mapping and profiles for different elements are shown in Fig. 68 and Fig. 69. Al

content in AlGaN calculated by atomic percentage is 81% (±2%) which agrees well with

XRD fitting. B content in BAlN is estimated to be 15% (±3%).

AFM image is shown in Fig. 70. The value of room mean square (RMS) roughness

was calculated by scanning over 20 × 20 µm2 to get an average information. The RMS of

this DBR is 9.1 nm, and the interface roughness is assumed to be equal to surface rough-

ness. Since BAlN is polycrystalline, the layers are considered to be fully relaxed. The

effect of the strain is also pretty weak when compared to the roughness, as discussed in the

119

Figure 69: EDX profiles of different elements along the growth direction of 18-pairBAlN/AlGaN DBR structure reflecting at 265 nm with (a) lower and (b) higher magni-fication.

120

Figure 70: AFM image of 18-pair BAlN/AlGaN DBR reflecting at 265 nm.

subsection 5.3.2.

The experimental reflection spectrum and simulations with or without considering qual-

ity factors are shown in Fig. 71. The simulations used the compositions and thicknesses

determined from characterizations. The theoretical central wavelength of reflection agrees

well with the experimental data. The reflection of 70% was obtained experimentally, but

the theoretical value for the perfect structure reaches 95%. The quality issues cause this

deviation. By considering interface roughness and surface roughness, the reflection peak

drops to 75%. The deviation between experimental data and simulation is around 7%.

According to the simulations, the roughness is the main reason which limits the re-

flectivity. Further studies will be focused on the optimization of BAlN layers to reduce

roughness, for example by reducing the boron content in BAlN layers or the possible pla-

narization method. The high quality AlN layers, AlGaN layers with more than 70% Al and

BAlN layers with boron content less than 5% will be investigated. The plans are proposed

in the perspective section in the last chapter.

121

Figure 71: Comparison between experimental reflection spectrum and simulations of 18-pair BAlN/AlGaN DBR reflecting at 265 nm.

122

CHAPTER 6

CONCLUSION AND PERSPECTIVE

6.1 Conclusion

Semiconductor light sources emitting in the deep ultra-violet (DUV) region have numer-

ous applications such as sterilization, water/air purification, optical imaging systems, spec-

troscopy, and high density storage systems. III-nitrides based VCSELs have raised more

and more interests in the recent years for the promising properties. However, the realization

of VCSELs operating in DUV range needs to satisfy very critical requirements: high emis-

sion efficiency of MQWs and highly reflective DBRs which are extremely difficult for the

short wavelengths. It means that the AlN substrates should be in good quality with disloca-

tion density less than 108 cm−2, QW emission should be TE-polarization dominant (surface

emission), two materials constructing DBRs should have high refractive index contrast,

and DBR structures should have minimum defects and roughness at the same time. Each

of them would be an arduous task.

VCSEL structure has two main blocks: MQWs and DBRs. This work is focused on

the MOVPE growth and studies of these two parts for the development of final devices,

which includes the growth and characterizations of BAlGaN materials, AlGaN-based DUV

MQWs with enhanced TE-polarized emission, and DBRs based on novel BAlN/AlGaN

material system.

For AlGaN material, a careful control of composition and relaxation for AlGaN MOVPE

growth has been explored. The composition pulling effect in AlGaN grown on AlN tem-

plates has been observed and investigated. The critical thickness range of AlGaN under

compressive strain grown on AlN is estimated by introducing different methods reported in

the literature. Theoretical curves and experimental values have been combined in a chart

which can be used as a reference for the experiments. Threading dislocation densities have

been calculated based on the XRD measurements and it’s an efficient way to quantify the

123

quality of the layers for comparison without requiring cross-sectional TEM images.

MOVPE growth of BAlN alloys with high boron incorporation has been achieved. The

structural and optical properties of 5-period AlN/BAlN heterostructure with 11% boron

grown at 1000 ◦C has been analyzed. Then, growth conditions of low temperature with

annealing has been investigated. It was found that low temperature has alleviated boron

poisoning during growth under high TEB/III ratio and improved crystallinity to obtain clear

XRD peaks corresponding to BAlN, which are very original results concerning this new

material. Wurtzite BAlN layers containing as high as 12% boron grown on both GaN and

AlN templates exhibit clear XRD peaks. The results for the growth of BAlN material and

the study of characteristics can advance prospects for MOVPE-grown boron alloys, which

can result in more freedom in bandgap, strain engineering and refractive index engineering

for eventual deep-UV sources.

For the study of the active region, a 4-period Al0.57Ga0.43N / Al0.38Ga0.62N MQWs struc-

ture has been grown on a relaxed Al0.58Ga0.42N buffer on AlN templates. The composition

of the quantum wells was optimized so that the strain present in wells is sufficient to en-

hance TE-polarized (E-field ⊥ c) emission. The relaxed AlGaN buffer on AlN template

serves as pseudo-substrate, and in this way the barriers are almost strain-free which limits

the formation of strain-related defects in the quantum wells. The structure exhibits an emis-

sion peak at 286 nm with a sharp linewidth at 77 K. Transmission measurements combined

with simulations confirm a sufficient oscillator strength leading to an optical absorption

coefficient in the wells as high as 3×105 cm−1. The results represent an important step to-

wards the development of DUV light sources, especially surface-emitting LEDs and lasers.

Based on the results of 4-period MQWs, 10- and 20-period MQWs have been grown and

characterized, which will be used for the processing of light emission devices by depositing

dielectric DBR on the top and at the bottom. Besides, the threading dislocations and V-pits

in the QW samples have been characterized and their origin is discussed. The influence of

V-pits on the structural quality of the MQWs and on optical emission at 280 nm has also

124

been analyzed. It has been observed that near-surface V-pits were always associated with

grain boundaries consisting of edge threading dislocations originating from the AlN/Al2O3

interface. Although the high density of V-pits disrupted MQWs growth, it did not affect the

internal quantum efficiency which was measured to be ∼1% at room temperature even when

V-pit density was increased by 30 times. The results help to understand the origin, propa-

gation and effects of the typical defects in AlGaN MQWs grown on AlN/Al2O3 templates

which may lead to further improvement of the performance of DUV devices.

The last part is for the simulation and realization of DUV DBRs. The simulation soft-

ware based on transfer-matrix methods was developed before in our group, and it assumes

that the structure is perfect without any quality degradation. In this work, different quality

factors such as surface roughness, interface roughness and strain have been introduced into

the simulation and their influence on the performance of DBRs have been discussed. In

terms of experiments, different growth conditions for BAlGaN DBRs have been compared

and discussed. 70% reflection has been achieved in the DUV region at 260 nm and at 280

nm with stopband width of 19 and 18 nm by using 18-pair BAlN/Al(Ga)N structures. The

experimental data are comparable with the simulations considering quality factors. Al-

though it still requires more effort to improve structural quality and surface roughness, the

progress achieved can help to develop final DBRs with reflection more than 90% and to

apply novel BAlN/AlGaN DBRs for deep UV RC-LEDs and VCSELs.

In summary, the work of this thesis has made progress on the aspects of MOVPE-grown

BAlGaN materials, MQWs emitting at 280 nm and optical investigations, simulation and

realization of BAlN/AlGaN DBRs. The work aims at the final DUV devices, and the results

obtained can also be helpful and informative for other applications of BAlGaN materials.

For the target to realize DUV light sources, on one hand, 10-period and 20-period MQW

samples emitting at 280 nm are in the process of depositing highly reflective dielectric

DBRs on the top and at the bottom for the vertical cavity. The first tests will be performed

after that. At the same time, a new Aixtron 3×2 inch, close coupled showerhead (CCS)

125

MOVPE system has been brought into operation, and the substrate temperature can reach

1300 ◦C. The future research work has been proposed in the next section.

6.2 Perspective

The perspective of this work covers further improvement of MQW and DBR performances,

as well as the processing for final devices. The plans are outlined in this section.

• Improvement of AlN templates

The key factor for increasing IQE of DUV MQWs is to reduce dislocation density of AlN

substrates. AlN native bulk substrates can be one choice, but it is not commercially avail-

able. So it’s necessary to grow and optimize AlN/sapphire templates with low dislocation

densities and low roughness. The new CCS reactor makes it possible, with which the sub-

strate surface can be heated up to 1300 ◦C. The growth temperatures higher than 1100 ◦C

can help to increase the diffusion length of Al atoms and decrease its affinity to oxygen.

Figure 72 shows the first results of AlN layers grown on the sapphire substrates in the CCS

reactor. 1 µm AlN layer shows smooth surface with RMS of 0.47 nm and growth steps can

be observed by AFM images. There are no cracks over the 2 inch sample surface. The

full-widths at half-maximum (FWHMs) are 757 arcsec for (0 0 0 2) planes, 1065 arcsec

for (1 0 -1 5) planes and 1703 arcsec for (1 0 -1 2) planes. The dislocation densities were

estimated to be 1.2 × 109 cm−2 for the screw component and 3.1 × 1010 cm−2 for the edge

component. The parameters such as strain releasing layers, V/III ratio, temperature and

thickness will be adjusted for the further improvement.

• Improvement of MQW efficiency

Besides the high quality substrates, other schemes can be used to enhance IQE. For exam-

ple, by doping In, MQWs might get IQE as high as 80% despite high threading dislocation

density [181, 182]. Si doping can also be an option to enhance the emission from MQWs,

since Si can help Al migration and then improve the quality of the MQWs [48, 183]. In the

126

Figure 72: (a) SEM image and (b) AFM image of AlN template grown on sapphire in CCSreactor.

literature, the IQE of MQWs for the wavelength of 288 nm was increased from 29.7% to

42% by doping Si [48].

• Optimization of DBRs

Based on the optimized AlN layers, the next steps would be the optimization of AlGaN

containing more than 70% Al with minimum pit density and monocrystalline BAlN con-

taining less than 5% B. As explained in Section 3.2.2, low temperature can help to alleviate

boron poisoning issue for high boron content. However, for low boron content, the condi-

tions are more close to the AlN conditions and the temperature higher than 1100 ◦C might

be favored. By reducing defect density and roughness of both AlGaN and BAlN layers, the

performance of BAlN/AlGaN DBRs would be enhanced.

• Processing of devices

For the device structures, there would be several alternative plans. AlGaN MQWs sand-

wiched by dielectric DBRs is under processing right now, and the structure is shown in Fig.

73(a). After optimizing BAlN/AlGaN DBRs, the following AlGaN active region will be

grown and then dielectric DBR will be deposited on the top for the tests of RC-LED and

127

Figure 73: (a) VCSEL design with dielectric top and bottom mirrors; (b) VCSEL designwith dielectric top mirror and BAlN/AlGaN bottom mirror; (c) inverted VCSEL structurewith BAlN/AlGaN bottom mirror grown on the active region.

of VCSEL structures, as shown in Fig. 73(b). If the surface of BAlN/AlGaN is not smooth

enough for the following growth of AlGaN active region, the inverted structure can be con-

sidered: active region grown directly on the AlN template at first and then BAlN/AlGaN

DBR on top of it, as shown in Fig. 73(c). In this case, the cavity would be extended in the

substrate.

6.3 Publications and awards

Peer-reviewed articles:

1. X. Li, G. Le Gac, S. Bouchoule, Y. E. Gmili, G. Patriarche, S. Sundaram, P. Disseix,

F. Reveret, J. Leymarie, J. Streque, F. Genty, J.-P. Salvestrini, R. D. Dupuis, X.-H.

128

Li, P. L. Voss, and A. Ougazzaden, “Structural and optical investigations of AlGaN

MQWs grown on a relaxed AlGaN buffer on AlN templates emitting at 280 nm,”

Journal of Crystal Growth, 432: 37-44, 2015.

2. X. Li, S. Sundaram, P. Disseix, G. Le Gac, S. Bouchoule, G. Patriarche, F. Reveret,

J. Leymarie, Y. E. Gmili, T. Moudakir, F. Genty, J.-P. Salvestrini, R. D. Dupuis, P. L.

Voss, and A. Ougazzaden, “AlGaN-based MQWs grown on a thick relaxed AlGaN

buffer on AlN templates emitting at 285 nm,” Optical Materials Express, 5(2): 380-

392, 2015.

3. X. Li, S. Sundaram, Y. El Gmili, T. Moudakir, F. Genty, S. Bouchoule, G. Patriache,

R. Dupuis, P. Voss, J.-P. Salvestrini, and A. Ougazzaden, “BAlN thin layers for deep

UV applications,” Physica Status Solidi (a), 212(4): 745-750, 2015.

4. X. Li, S. Sundaram, Y. El Gmili, F. Genty, S. Bouchoule, G. Patriache, P. Disseix,

F. Reveret, J. Leymarie, J.-P. Salvestrini, R. Dupuis, P. Voss, and A. Ougazzaden,

“MOVPE grown periodic AlN/BAlN heterostructure with high boron content,” Jour-

nal of Crystal Growth, 414: 119-122, 2015.

5. S. Sundaram, R. Puybaret, Y. El Gmili, X. Li, P. L. Bonanno, K. Pantzas, G. Orsal, Z.-

H. Cai, G. Patriarche, P. L. Voss, J. P. Salvestrini, and A. Ougazzaden, “Nanoselective

area growth and in-depth characterization of dislocation-free InGaN nanopyramids

on AlN buffered Si(111) templates,” Applied Physics Letters, 107: 113105, 2015.

6. S. Sundaram, R. Puybaret, X. Li, Y. El Gmili, J. Streque, K. Pantzas, G. Orsal, G. Pa-

triarche, P. L. Voss, J. P. Salvestrini, and A. Ougazzaden, “High quality thick InGaN

nanostructures grown by nanoselective area growth for new generation photovoltaic

devices,” Physica Status Solidi (a), 212(4): 740-744, 2015.

7. S. Sundaram, R. Puybaret, Y. El Gmili, X. Li, P. L. Bonanno, K. Pantzas, G. Orsal,

129

D. Troadec, Z.-H. Cai, G. Patriarche, P. L. Voss, J. P. Salvestrini, and A. Ougaz-

zaden, “Nanoscale selective area growth of thick, dense, uniform, In-rich, InGaN

nanostructure arrays on GaN/sapphire template,” Journal of Applied physics, 116:

163105, 2014.

Awards:

1. Doctoral Consortium on Photonics, Optics and Lasers Technology 2015 (DCPHO-

TOPTICS 2015) (Berlin, Germany) - Best Paper Award, 13th March, 2015.

2. European Materials Research Society (E-MRS) 2014 Spring meeting (Lille, France)

- Graduate Student Award, 30th May, 2014.

Communications:

1. X. Li, S. Sundaram, Y. El Gmili, S. Bouchoule, G. Patriarche, F. Genty, J-P. Salvestrini,

R. D. Dupuis, P.L. Voss, and A. Ougazzaden. ”MOVPE growth and characterizations

of novel BAlN/AlGaN Bragg mirrors reflecting at 265 nm,” 11th International Con-

ference on Nitride Semiconductors (ICNS-11), Beijing, China, 30th August - 4th

September 2015. (poster)

2. X. Li, Y. El Gmili, G. Le Gac, S. Sundaram, S. Bouchoule, G. Patriarche, P. Disseix,

F. Reveret, J. Leymarie, J. Streque, F. Genty, J-P. Salvestrini, R. D. Dupuis, P.L. Voss,

and A. Ougazzaden. “Defects in AlGaN MQWs grown on AlN templates and optical

investigations for emission at 280 nm,” The European Workshop on Metalorganic

Vapour Phase Epitaxy (EWMOVPE 2015), Lund, Sweden, 6th - 10th June 2015.

(poster)

3. X. Li, S. Sundaram, P. Disseix, S. Bouchoule, G. Le Gac, G. Patriarche, F. Reveret, J.

Leymarie, Y. El Gmili, J. Streque, F. Genty, J-P. Salvestrini, P.L. Voss, R. D. Dupuis,

and A. Ougazzaden, “BAlGaN-based Vertical Cavity Surface Emitting Laser Oper-

ating in Deep UV Region,” Doctoral Consortium of 3rd International Conference on

130

Photonics, Optics and Laser Technology (DCPHOTOPICS2015), Berlin, Germany,

12th - 14th March, 2015. (oral presentation)

4. X. Li, S. Sundaram, Y. El Gmili, F. Genty, S. Bouchoule, G. Patriache, P. Disseix,

F. Reveret, J. Leymarie, J.-P. Salvestrini, R. Dupuis, P. Voss, and A. Ougazzaden,

“MOVPE grown periodic AlN/BAlN heterostructure with high boron content,” 17th

International Conference on Metalorganic Vapor Phase Epitaxy (ICMOVPE2014),

Lausanne, Switzerland, 13th - 18th July 2014. (poster)

5. X. Li, S. Sundaram, Y.El Gmili, T.Moudakir, P. Disseix, F.Reveret, J. Leymarie,

S. Bouchoule, F. Genty, J-P. Salvestrini, R. D. Dupuis, P.L. Voss, A. Ougazzaden,

“AlGaN-based multi-quantum wells emitting at 285 nm grown on a thick AlGaN

relaxed buffer on AlN template,” 17th International Conference on Metalorganic Va-

por Phase Epitaxy (ICMOVPE2014), Lausanne, Switzerland, 13th - 18th July 2014.

(poster)

6. X. Li, S. Sundaram, Y. El Gmili, T. Moudakir, F. Genty, S. Bouchoule, G. Patri-

ache, R. Dupuis, P. Voss, J.-P. Salvestrini, and A. Ougazzaden, “BAlN thin layers

for deep UV applications,” European materials society, 2014 SPRING MEETING

(E-MRS2014), Lille, France, 26th - 30th May 2014. (oral presentation)

131

REFERENCES

[1] “http://www.ioffe.ru/SVA/NSM/Semicond/.”

[2] “http://www.compoundsemiconductor.net.”

[3] “http://en.wikipedia.org/wiki/Wurtzite crystal structure.”

[4] G. Rakotonanahary, Spectroscopie des transitions excitoniques dans des puits quan-

tiques GaN/AlGaN. PhD thesis, Universite Blaise-Pascal-Clermont-Ferrand II, 2011.

[5] E. T. Yu, X. Z. Dang, P. M. Asbeck, S. S. Lau, G. J. Sullivan, and I. Introduction,

“Spontaneous and piezoelectric polarization effects in III-V nitride heterostructures,”

Journal of Vacuum Science & Technology B, vol. 17, no. 4, pp. 1742–1749, 1999.

[6] R. People and J. C. Bean, “Calculation of critical layer thickness versus lattice

mismatch for GexSi1−x/Si strained-layer heterostructures,” Applied Physics Letters,

vol. 47, no. 3, pp. 322–324, 1985.

[7] A. Fischer, H. Kuhne, and H. Richter, “New approach in equilibrium theory for

strained layer relaxations,” Physical Review Letters, vol. 73, no. 20, pp. 2712–2715,

1994.

[8] M. Abid, T. Moudakir, Z. Djebbour, G. Orsal, S. Gautier, A. En Naciri, A. Migan-

Dubois, and A. Ougazzaden, “Blue-violet boron-based Distributed Bragg Reflectors

for VCSEL application,” Journal of Crystal Growth, vol. 315, pp. 283–287, 2011.

[9] M. ABID, Design and epitaxial growth of vertical-cavity surface-emitting lasers

(VCSEL) emitting at ultraviolet wavelength. PhD thesis, Georgia Institute of Tech-

nology, 2013.

[10] D. Brunner, H. Angerer, E. Bustarret, F. Freudenberg, R. Hopler, R. Dimitrov,

O. Ambacher, and M. Stutzmann, “Optical constants of epitaxial AlGaN films

132

and their temperature dependence,” Journal of Applied Physics, vol. 82, no. 1997,

pp. 5090–5096, 1997.

[11] “http://en.wikipedia.org/wiki/Ultraviolet germicidal irradiation.”

[12] “http://en.wikipedia.org/wiki/Ultraviolet light therapy.”

[13] “http://en.wikipedia.org/wiki/Ultraviolet photography.”

[14] “http://en.wikipedia.org/wiki/Fluorescence.”

[15] “http://en.wikipedia.org/wiki/UV curing.”

[16] “http://en.wikipedia.org/wiki/Optical storage.”

[17] “http://en.wikipedia.org/wiki/Ultraviolet photography.”

[18] H. Hirayama, J. Norimatsu, N. Noguchi, S. Fujikawa, T. Takano, K. Tsubaki, and

N. Kamata, “Milliwatt power 270 nm-band AlGaN deep-UV LEDs fabricated on

ELO-AlN templates,” Physica Status Solidi (c), vol. 6, no. S2, pp. S474–S477, 2009.

[19] K. Nagamatsu, N. Okada, H. Sugimura, H. Tsuzuki, F. Mori, K. Iida, A. Bando,

M. Iwaya, S. Kamiyama, H. Amano, and I. Akasaki, “High-efficiency AlGaN-based

UV light-emitting diode on laterally overgrown AlN,” Journal of Crystal Growth,

vol. 310, pp. 2326–2329, 2008.

[20] A. Knauer, V. Kueller, U. Zeimer, M. Weyers, C. Reich, and M. Kneissl, “AlGaN

layer structures for deep UV emitters on laterally overgrown AlN/sapphire tem-

plates,” Physica Status Solidi (a), vol. 210, no. 3, pp. 451–454, 2013.

[21] M. Martens, F. Mehnke, C. Kuhn, C. Reich, V. Kueller, A. Knauer, C. Netzel,

C. Hartmann, J. Wollweber, J. Rass, T. Wernicke, M. Bickermann, M. Weyers, and

M. Kneissl, “Performance characteristics of UV-C AlGaN-Based lasers grown on

133

sapphire and bulk AlN substrates,” IEEE Photonics Technology Letters, vol. 26,

no. 4, pp. 342–345, 2014.

[22] M. E. Hawkridge, Z. Liliental-Weber, H. J. Kim, S. Choi, D. Yoo, J.-H. Ryou, and

R. D. Dupuis, “Erratic dislocations within funnel defects in AlN templates for Al-

GaN epitaxial layer growth,” Applied Physics Letters, vol. 94, p. 171912, 2009.

[23] L. W. Sang, H. Fang, Z. X. Qin, X. Q. Wang, B. Shen, Z. J. Yang, G. Y. Zhang, X. P.

Zhang, L. P. You, and D. P. Yu, “Transmission electron microscopy investigation

of inversion domain boundary in Al0.65Ga0.35N grown on AlN/sapphire template,”

Applied Physics Letters, vol. 95, p. 112106, 2009.

[24] F. Mehnke, T. Wernicke, H. Pingel, C. Kuhn, C. Reich, V. Kueller, A. Knauer,

M. Lapeyrade, M. Weyers, and M. Kneissl, “Highly conductive n-AlxGa1−xN lay-

ers with aluminum mole fractions above 80%,” Applied Physics Letters, vol. 103,

p. 212209, 2013.

[25] M. L. Nakarmi, K. H. Kim, K. Zhu, J. Y. Lin, and H. X. Jiang, “Transport properties

of highly conductive n-type Al-rich AlxGa1−xN (x≥0.7),” Applied Physics Letters,

vol. 85, no. 17, pp. 3769–3711, 2004.

[26] R. Collazo, S. Mita, J. Xie, A. Rice, J. Tweedie, R. Dalmau, and Z. Sitar, “Progress

on n-type doping of algan alloys on aln single crystal substrates for uv optoelectronic

applications,” Physica Status Solidi (c), vol. 8, no. 7-8, pp. 2031–2033, 2011.

[27] D. Nilsson, X. T. Trinh, E. Janzen, N. T. Son, and A. Kakanakova-Georgieva, “On

the behavior of silicon donor in conductive AlxGa1−xN (0.63 ≤ x ≤ 1),” Physica

Status Solidi (b), vol. 252, no. 6, pp. 1306–1310, 2015.

[28] T. Kinoshita, T. Obata, H. Yanagi, and S. I. Inoue, “High p-type conduction in high-

Al content Mg-doped AlGaN,” Applied Physics Letters, vol. 102, p. 012105, 2013.

134

[29] J. Ding, H. Jeon, a. V. Nurmikko, H. Luo, N. Samarth, and J. K. Furdyna, “Laser

action in the blue-green from optically pumped (Zn,Cd)Se/ZnSe single quantum well

structures,” Applied Physics Letters, vol. 57, no. 26, pp. 2756–2758, 1990.

[30] B. Gil, Group III nitride semiconductor compounds. Series on semiconductor sci-

ence and technology: 6, Clarendon Press, 1998.

[31] F. G. McIntosh, K. S. Boutros, J. C. Roberts, S. M. Bedair, E. L. Piner, and N. A.

El-Masry, “Growth and characterization of AlInGaN quaternary alloys,” Applied

Physics Letters, vol. 68, no. 1, pp. 40–42, 1996.

[32] M. Shatalov, J. Zhang, A. S. Chitnis, S. Member, V. Adivarahan, J. Yang, G. Simin,

and M. A. Khan, “Deep ultraviolet light-emitting diodes using quaternary AlInGaN

multiple quantum wells,” IEEE Journal of Selected Topics in Quantum Electronics,

vol. 8, no. 2, pp. 302–309, 2002.

[33] “http://www.nobelprize.org/nobel prizes/physics/laureates/2014/.”

[34] I. Akasaki, H. Amano, S. Sota, H. Sakai, T. Tanaka, and M. Koike, “Stimulated

emission by current injection from an AlGaN/GaN/GaInN quantum well device,”

Japanese Journal of Applied Physics, vol. 34, pp. 1517–1519, 1995.

[35] J. Han, M. H. Crawford, R. J. Shul, J. J. Figiel, M. Banas, L. Zhang, Y. K. Song,

H. Zhou, and A. V. Nurmikko, “AlGaN/GaN quantum well ultraviolet light emitting

diodes,” Applied Physics Letters, vol. 73, no. 12, pp. 1688–1690, 1998.

[36] M. Shatalov, W. Sun, A. Lunev, X. Hu, A. Dobrinsky, Y. Bilenko, J. Yang, M. Shur,

R. Gaska, C. Moe, G. Garrett, and M. Wraback, “AlGaN deep-ultraviolet light-

emitting diodes with external quantum efficiency above 10%,” Applied Physics Ex-

press, vol. 5, p. 082101, 2012.

135

[37] M. Ippommatsu, A. Hirano, I. Akasaki, and H. Amano, “Development of AlGaN

DUV-LED,” in 2013 Conference on Lasers and Electro-Optics Pacific Rim, (Optical

Society of America, 2013), paper MH1 3, 2013.

[38] M. Kneissl, Z. Yang, M. Teepe, C. Knollenberg, O. Schmidt, P. Kiesel, N. M. John-

son, S. Schujman, and L. J. Schowalter, “Ultraviolet semiconductor laser diodes on

bulk AlN,” Journal of Applied Physics, vol. 101, p. 123103, 2007.

[39] H. Tsuzuki, F. Mori, K. Takeda, M. Iwaya, S. Kamiyama, H. Amano, I. Akasaki,

H. Yoshida, M. Kuwabara, Y. Yamashita, and H. Kan, “Novel UV devices on high-

quality AlGaN using grooved underlying layer,” Journal of Crystal Growth, vol. 311,

no. 10, pp. 2860–2863, 2009.

[40] H. Kim-Chauveau, E. Frayssinet, B. Damilano, P. De Mierry, L. Bodiou, L. Nguyen,

P. Vennegues, J.-M. Chauveau, Y. Cordier, J. Duboz, R. Charash, A. Vajpeyi, J.-M.

Lamy, M. Akhter, P. Maaskant, B. Corbett, A. Hangleiter, and A. Wieck, “Growth

optimization and characterization of lattice-matched Al0.82In0.18N optical confine-

ment layer for edge emitting nitride laser diodes,” Journal of Crystal Growth,

vol. 338, no. 1, pp. 20–29, 2012.

[41] H. Yoshida, M. Kuwabara, Y. Yamashita, K. Uchiyama, and H. Kan, “The current

status of ultraviolet laser diodes,” Physica Status Solidi (a), vol. 208, no. 7, pp. 1586–

1589, 2011.

[42] K. H. Li, X. Liu, Q. Wang, S. Zhao, and Z. Mi, “Ultralow-threshold electrically in-

jected AlGaN nanowire ultraviolet lasers on Si operating at low temperature,” Nature

Nanotechnology, vol. 10, pp. 140–144, 2015.

[43] J. Xie, S. Mita, Z. Bryan, W. Guo, L. Hussey, B. Moody, R. Schlesser, R. Kirste,

M. Gerhold, R. Collazo, and Z. Sitar, “Lasing and longitudinal cavity modes in

136

photo-pumped deep ultraviolet AlGaN heterostructures,” Applied Physics Letters,

vol. 102, p. 171102, 2013.

[44] N. M. Johnson, B. Cheng, S. Choi, C. L. Chua, C. Knollenberg, J. E. Northrup, M. R.

Teepe, T. Wunderer, and Yang Z., “Deep UV AlGaN lasers,” in the 9th International

Symposium on Semiconductor Light Emitting Devices, Berlin, 2012.

[45] H. Hirayama, N. Maeda, S. Fujikawa, S. Toyoda, and N. Kamata, “Recent progress

and future prospects of AlGaN-based high-efficiency deep-ultraviolet light-emitting

diodes,” Japanese Journal of Applied Physics, vol. 53, p. 100209, 2014.

[46] H. Hirayama, T. Yatabe, N. Noguchi, T. Ohashi, and N. Kamata, “231-261 nm

AlGaN deep-ultraviolet light-emitting diodes fabricated on AlN multilayer buffers

grown by ammonia pulse-flow method on sapphire,” Applied Physics Letters,

vol. 91, p. 071901, 2007.

[47] X.-H. Li, T. Detchprohm, T.-T. Kao, M. M. Satter, S.-C. Shen, P. Douglas Yoder,

R. D. Dupuis, S. Wang, Y. O. Wei, H. Xie, A. M. Fischer, F. A. Ponce, T. Wernicke,

C. Reich, M. Martens, and M. Kneissl, “Low-threshold stimulated emission at 249

nm and 256 nm from AlGaN-based multiple-quantum-well lasers grown on sapphire

substrates,” Applied Physics Letters, vol. 105, p. 141106, 2014.

[48] Y. Tian, J. Yan, Y. Zhang, X. Chen, Y. Guo, P. Cong, L. Sun, Q. Wang, E. Guo,

X. Wei, J. Wang, and J. Li, “Stimulated emission at 288 nm from silicon-doped

AlGaN-based multiple-quantum-well laser,” Optics Express, vol. 23, no. 9, p. 11334,

2015.

[49] S. Zhao, A. T. Connie, M. H. T. Dastjerdi, X. H. Kong, Q. Wang, M. Djavid, S. Sadaf,

X. D. Liu, I. Shih, H. Guo, and Z. Mi, “Aluminum nitride nanowire light emitting

diodes: Breaking the fundamental bottleneck of deep ultraviolet light sources,” Sci-

entific Reports, vol. 5, p. 8332, 2015.

137

[50] S. Zhao, X. Liu, S. Y. Woo, J. Kang, G. a. Botton, and Z. Mi, “An electrically

injected AlGaN nanowire laser operating in the ultraviolet-C band,” Applied Physics

Letters, vol. 107, p. 043101, 2015.

[51] J. M. Redwing, D. A. Loeber, N. G. Anderson, M. A. Tischler, and J. S. Flynn,

“An optically pumped GaN-AlGaN vertical cavity surface emitting laser,” Applied

Physics Letters, vol. 69, no. 1, pp. 1–3, 1996.

[52] I. L. Krestnikov and N. N. Ledentsov, “Photopumped InGaN /GaN/AlGaN Vertical

Cavity Surface Emitting Laser Operating at Room Temperature,” Physica Status

Solidi (b), vol. 216, pp. 511–516, 1999.

[53] Y. Higuchi, K. Omae, H. Matsumura, and T. Mukai, “Room-Temperature CW lasing

of a GaN-based vertical-cavity surface-emitting laser by current injection,” Applied

Physics Express, vol. 1, p. 121102, 2008.

[54] B. C. Lin, Y. A. Chang, K. J. Chen, C. H. Chiu, Z. Y. Li, Y. P. Lan, C. C. Lin, P. T.

Lee, Y. K. Kuo, M. H. Shih, H. C. Kuo, T. C. Lu, and S. C. Wang, “Design and

fabrication of a InGaN vertical-cavity surface-emitting laser with a composition-

graded electron-blocking layer,” Laser Physics Letters, vol. 11, p. 085002, 2014.

[55] D. Kasahara, D. Morita, T. Kosugi, K. Nakagawa, J. Kawamata, Y. Higuchi, H. Mat-

sumura, and T. Mukai, “Demonstration of Blue and Green GaN-Based Vertical-

Cavity Surface-Emitting Lasers by Current Injection at Room Temperature,” Applied

Physics Express, vol. 4, p. 072103, 2011.

[56] O. Mitrofanov, S. Schmult, M. J. Manfra, T. Siegrist, N. G. Weimann, A. M. Sergent,

and R. J. Molnar, “High-reflectivity ultraviolet AlGaN/AlGaN distributed Bragg re-

flectors,” Applied Physics Letters, vol. 88, p. 171101, 2006.

[57] E. Feltin, J. F. Carlin, J. Dorsaz, G. Christmann, R. Butt, M. Laugt, M. Ilegems, and

138

N. Grandjean, “Crack-free highly reflective AlInN/AlGaN Bragg mirrors for UV

applications,” Applied Physics Letters, vol. 88, p. 051108, 2006.

[58] C. G. Moe, Y. Wu, J. Piprek, S. Keller, J. S. Speck, S. P. DenBaars, and D. Emerson,

“AlGaN/AlN distributed bragg reflectors for deep ultraviolet wavelengths,” Physica

Status Solidi (a), vol. 203, no. 8, pp. 1915–1919, 2006.

[59] A. R. Getty, A. David, Y. Wu, C. Weisbuch, and J. S. Speck, “Demonstration of

Distributed Bragg Reflectors for Deep Ultraviolet Applications,” Japanese Journal

of Applied Physics, vol. 46, no. 32, pp. 767–769, 2007.

[60] L. Zhang, K. Dong, D. Chen, Y. Liu, J. Xue, H. Lu, R. Zhang, and Y. Zheng, “Solar-

blind ultraviolet AlInN/AlGaN distributed Bragg reflectors,” Applied Physics Let-

ters, vol. 102, p. 242112, 2013.

[61] L. Zhang, Z.-H. Liu, X.-G. Huang, Q.-F. LI, R. Zhang, Z.-L. Xie, and X.-Q. Xiu,

“Properties of an AlGaN/AlN distributed-Bragg-reflector structure,” Journal of Ko-

rean Physical Society, vol. 65, no. 7, pp. 1101–1105, 2014.

[62] M. P. Hoffmann, A. Franke, M. Bobea, F. Kaess, R. Kirste, A. Dadgar, A. Strittmat-

ter, M. Gerhold, R. Collazo, and Z. Sitar, “High reflective AlN/AlGaN distributed

Bragg reflectors for UV microcavity lasers,” Gallium Nitride Materials and Devices

II, Proc. of SPIE, vol. 6473, p. 64731G, 2015.

[63] S. Watanabe, T. Takano, K. Jinen, J. Yamamoto, and H. Kawanishi, “Refractive

indices of BxAl1−xN (x= 0-0.012) and ByGa1−yN (y= 0-0.023) epitaxial layers in

ultraviolet region,” Physica Status Solidi (c), vol. 0, no. 7, pp. 2691–2694, 2003.

[64] M. Abid, T. Moudakir, G. Orsal, S. Gautier, A. En Naciri, Z. Djebbour, J.-H. Ryou,

G. Patriarche, L. Largeau, H. J. Kim, Z. Lochner, K. Pantzas, D. Alamarguy, F. Jo-

mard, R. D. Dupuis, J.-P. Salvestrini, P. L. Voss, and A. Ougazzaden, “Distributed

139

Bragg reflectors based on diluted boron-based BAlN alloys for deep ultraviolet op-

toelectronic applications,” Applied Physics Letters, vol. 100, p. 051101, 2012.

[65] Y.kuga, T. Shirai, M. Haruyama, H. Kawanishi, and Y. Suematsu, “Violet and

Near-UV Light Emission from GaN/Al0.08Ga0.92N Injection Diode Grown on (0001)

6H-SiC Substrate by Low-Pressure Metal-Organic Vapor Phase Epitaxy,” Japanese

Journal of Applied Physics, vol. 34, pp. 4085–4086, 1995.

[66] A. Y. Polyakov, M. Shin, M. Skowronski, D. W. Greve, R. G. Wilson, A. V. Gov-

orkov, and R. M. Desrosiers, “Growth of GaBN ternary solutions by organometallic

vapor phase epitaxy,” Journal of Electronic Materials, vol. 26, no. 3, pp. 237–242,

1997.

[67] A. Nakajima, Y. Furukawa, H. Yokoya, and H. Yonezu, “Growth of BxAl1−xN layers

using decaborane on SiC substrates,” Journal of Crystal Growth, vol. 278, pp. 437–

442, 2005.

[68] A. Y. Polyakov, M. Shin, W. Qian, M. Skowronski, D. W. Greve, and R. G. Wilson,

“Growth of AlBN solid solutions by organometallic vapor-phase epitaxy,” Journal

of Applied Physics, vol. 81, no. 4, pp. 1715–1719, 1997.

[69] M. Shibata, M. Kurimoto, J. Yamamoto, T. Honda, and H. Kawanishi, “GaN/BAlN

heterostructure grown on a (0001) 6H-SiC substrate by metalorganic vapor phase

epitaxy,” Journal of Crystal Growth, vol. 189-190, pp. 445–447, 1998.

[70] C. H. Wei, Z. Y. Xie, J. H. Edgar, K. C. Zeng, J. Y. Lin, H. X. Jiang, J. Chaud-

huri, C. Ignatiev, and D. N. Braski, “MOCVD growth of GaBN on 6H-SiC (0001)

substrates,” Journal of Electronic Materials, vol. 29, no. 4, pp. 452–456, 2000.

[71] T. Takano, M. Kurimoto, J. Yamamoto, and H. Kawanishi, “Epitaxial growth of

140

high quality BAlGaN quaternary lattice matched to AlN on 6H-SiC substrate by LP-

MOVPE for deep-UV emission,” Journal of Crystal Growth, vol. 237-239, pp. 972–

977, 2002.

[72] T. Akasaka and T. Makimoto, “Flow-rate modulation epitaxy of wurtzite AlBN,”

Applied Physics Letters, vol. 88, p. 041902, 2006.

[73] S. Gautier, C. Sartel, S. Ould Saad Hamady, N. Maloufi, J. Martin, F. Jomard, and

A. Ougazzaden, “MOVPE growth study of BxGa(1−x)N on GaN template substrate,”

Superlattices and Microstructures, vol. 40, pp. 233–248, 2006.

[74] A. Ougazzaden, S. Gautier, T. Moudakir, Z. Djebbour, Z. Lochner, S. Choi, H. J.

Kim, J.-H. Ryou, R. D. Dupuis, and A. A. Sirenko, “Bandgap bowing in BGaN thin

films,” Applied Physics Letters, vol. 93, p. 083118, 2008.

[75] H. Kawanishi and T. Honda, “Fabrications of BAlGaInN multi-Layer or quantum

well structure for blue to UV Light emitter and its applications to optical devices,”

Photonics Based on Wavelength Integration and Manipulation, vol. 2, pp. 19–24,

2005.

[76] J. Edgar, D. Smith, C. Eddy, C. Carosella, and B. Sartwell, “c-Boron-aluminum

nitride alloys prepared by ion-beam assisted deposition,” Thin Solid Films, vol. 298,

no. 1-2, pp. 33–38, 1997.

[77] V. K. Gupta, C. C. Wamsley, M. W. Koch, and G. W. Wicks, “Molecular beam epi-

taxy growth of boron-containing nitrides,” Journal of Vacuum Science & Technology

B: Microelectronics and Nanometer Structures, vol. 17, no. 3, pp. 1246–1248, 1999.

[78] L. Liljeholm, M. Junaid, T. Kubart, J. Birch, L. Hultman, and I. Katardjiev, “Synthe-

sis and characterization of (0001)-textured wurtzite Al1−xBxN thin films,” Surface

and Coatings Technology, vol. 206, no. 6, pp. 1033–1036, 2011.

141

[79] J.-H. Song, J.-L. Huang, J. C. Sung, S.-C. Wang, H.-H. Lu, and D.-F. Lii, “Interfacial

microstructure evolution of (B, Al)N films grown on diamond substrates,” Thin Solid

Films, vol. 519, no. 13, pp. 4212–4215, 2011.

[80] C. Wei and J. Edgar, “Unstable composition region in the wurtzite B1−x−yGaxAlyN

system,” Journal of Crystal Growth, vol. 208, pp. 179–182, 2000.

[81] T. Akasaka, Y. Kobayashi, and T. Makimoto, “Nonpolar AlBN (1120) and (1100)

films grown on SiC substrates,” Applied Physics Letters, vol. 91, p. 041914, 2007.

[82] A. R. Denton and N. W. Ashcroft, “Vegards law,” Physical Review A, vol. 43, no. 6,

pp. 3161–3164, 1991.

[83] S. Nakamura, “GaN Growth Using GaN Buffer Layer,” Japanese Journal of Applied

Physics, vol. 30, no. Part 2, No. 10A, p. L1705, 1991.

[84] S. R. Lee, A. F. Wright, M. H. Crawford, G. a. Petersen, J. Han, and R. M. Biefeld,

“The band-gap bowing of AlxGa1−x N alloys,” Applied Physics Letters, vol. 74,

no. 22, pp. 3344–3346, 1999.

[85] W. Shan, J. W. Ager, W. Walukiewicz, E. E. Haller, B. D. Little, J. J. Song, M. Schur-

man, Z. C. Feng, R. A. Stall, and B. Goldenberg, “Near-band-edge photolumines-

cence emission in AlxGa1−xN under high pressure,” Applied Physics Letters, vol. 72,

pp. 2274–2276, 1998.

[86] S. Azzi, A. Zaoui, and M. Ferhat, “On the importance of the band gap bowing in

Boron-based III¢V ternary alloys,” Solid State Communications, vol. 144, pp. 245–

248, 2007.

[87] I. Vurgaftman, J. R. Meyer, and L. R. Ram-Mohan, “Band parameters for III-V

compound semiconductors and their alloys,” Journal of Applied Physics, vol. 89,

no. 11, pp. 5815–5875, 2001.

142

[88] K. B. Nam, J. Li, M. L. Nakarmi, J. Y. Lin, and H. X. Jiang, “Unique optical prop-

erties of AlGaN alloys and related ultraviolet emitters,” Applied Physics Letters,

vol. 84, no. 25, pp. 5264–5266, 2004.

[89] M. Leroux, F. Semond, F. Natali, D. Byrne, F. Cadoret, B. Damilano, A. Dussaigne,

N. Grandjean, A. Le Louarn, S. Vezian, and J. Massies, “About some optical prop-

erties of AlxGa1−xN /GaN quantum wells grown by molecular beam epitaxy,” Super-

lattices and Microstructures, vol. 36, pp. 659–674, 2004.

[90] H.-Y. Ryu, I.-G. Choi, H.-S. Choi, and J.-I. Shim, “Investigation of light extrac-

tion efficiency in AlGaN deep-ultraviolet light-emitting diodes,” Applied Physics

Express, vol. 6, p. 062101, 2013.

[91] T. Kolbe, A. Knauer, C. Chua, Z. Yang, S. Einfeldt, P. Vogt, N. M. Johnson, M. Wey-

ers, and M. Kneissl, “Optical polarization characteristics of ultraviolet (In)(Al)GaN

multiple quantum well light emitting diodes,” Applied Physics Letters, vol. 97,

p. 171105, 2010.

[92] M. Leroux, S. Dalmasso, F. Natali, S. Helin, C. Touzi, S. Laugt, M. Passerel,

F. Omnes, F. Semond, J. Massies, and P. Gibart, “Optical characterization of

AlxGa1−xN alloys ( x<0.7 ) grown on sapphire or silicon,” Physica Status Solidi

(b), vol. 234, no. 3, pp. 887–891, 2002.

[93] F. Bernadini, V. Fiorentini, and D. Vanderbilt, “Spontaneous polarization and piezo-

electric constants of III-V nitrides,” Physical Revew B, vol. 56, pp. 10024–10027,

1997.

[94] O. Ambacher, “Growth and applications of Group III-nitrides,” Journal of Physics

D: Applied Physics, vol. 31, pp. 2653–2710, 1998.

[95] H. M. Ng, R. Harel, S. N. G. Chu, and A. Y. Cho, “The effect of built-in electric field

143

in GaN / AlGaN quantum wells with high AlN mole fraction,” Journal of Electronic

Materials, vol. 30, no. 3, pp. 134–137, 2001.

[96] V. Fiorentini, F. Bernardini, F. D. Sala, A. D. Carlo, and P. Lugli, “Effects

of macroscopic-polarization built-in electrostatic fields in III-V nitrides multi-

quantum-wells,” Physical Revew B, vol. 60, no. 12, pp. 8849–8858, 1999.

[97] K. Tsubouchi and N. Mikoshib, “Zero-Temperature-Coefficient SAW Devices on

A1N Epitaxial Films,” IEEE Transactions on Sonics and Ultrasonics, vol. SU-32,

pp. 634–644, 1985.

[98] A. Barker and M. Ilegems, “Infrared lattice vibrations and free-electron dispersion

in GaN,” Physical Revew B, vol. 7, no. 2, pp. 734–750, 1973.

[99] V. W. L. Chin, T. L. Tansley, and T. Osotchan, “Electron mobilities in gallium, in-

dium and aluminum nitrides,” Journal of Applied Physics, vol. 75, no. 11, pp. 7365–

7372, 1994.

[100] F. Bernardini, V. Fiorentini, and D. Vanderbilt, “Accurate calculation of polarization-

related quantities in semiconductors,” Physical Review B, vol. 63, p. 193201, 2001.

[101] N. Grandjean, B. Damilano, S. Dalmasso, M. Leroux, M. Laugt, and J. Massies,

“Built-in electric-field effects in wurtzite AlGaN/GaN quantum wells,” Journal of

Applied Physics, vol. 86, no. 7, pp. 3714–3720, 1999.

[102] “http://en.wikipedia.org/wiki/Quantum-confined Stark effect.”

[103] G. B. Stringfellow, Organometallic Vapor-Phase Epitaxy. Academic Press, 1999.

[104] A. Mircea, A. Ougazzaden, and R. Mellet, “Very uniform epitaxy,” Progress in Crys-

tal Growth and Characterization of Materials, vol. 19, no. 1-2, pp. 39–49, 1989.

144

[105] S. Gautier, C. Sartel, S. Ould-Saad, J. Martin, A. Sirenko, and A. Ougazzaden, “GaN

materials growth by MOVPE in a new-design reactor using DMHy and NH3,” Jour-

nal of Crystal Growth, vol. 298, pp. 428–432, 2007.

[106] M. A. Moram and M. E. Vickers, “X-ray diffraction of III-nitrides,” Reports on

Progress in Physics, vol. 72, p. 036502, 2009.

[107] “http://en.wikipedia.org/wiki/Secondary ion mass spectrometry.”

[108] “http://www.probion.fr/fr/86-accueil/139-physical-description-of-sims-analysis.”

[109] “http://en.wikipedia.org/wiki/Scanning transmission electron microscopy.”

[110] K. Pantzas, G. Patriarche, D. Troadec, S. Gautier, T. Moudakir, S. Suresh,

L. Largeau, O. Mauguin, P. L. Voss, and A. Ougazzaden, “Nanometer-scale, quanti-

tative composition mappings of InGaN layers from a combination of scanning trans-

mission electron microscopy and energy dispersive x-ray spectroscopy,” Nanotech-

nology, vol. 23, p. 455707, 2012.

[111] “http://en.wikipedia.org/wiki/Energy-dispersive X-ray spectroscopy.”

[112] “http://en.wikipedia.org/wiki/Atomic force microscopy.”

[113] Veeco Instruments Inc., “A Practical Guide to Scanning Probe Microscopy (SPM),”

tech. rep., 2005.

[114] “http://en.wikipedia.org/wiki/Scanning electron microscope.”

[115] “http://en.wikipedia.org/wiki/Cathodoluminescence.”

[116] “http://en.wikipedia.org/wiki/Photoluminescence.”

[117] “http://en.wikipedia.org/wiki/Fourier transform infrared spectroscopy.”

145

[118] X.-H. Li, Y. O. Wei, S. Wang, H. Xie, T.-T. Kao, M. M. Satter, S.-C. Shen, P. Douglas

Yoder, T. Detchprohm, R. D. Dupuis, A. M. Fischer, and F. A. Ponce, “Temperature

dependence of the crystalline quality of AlN layer grown on sapphire substrates

by metalorganic chemical vapor deposition,” Journal of Crystal Growth, vol. 414,

pp. 76–80, 2015.

[119] X.-H. Li, S. Wang, H. Xie, Y. O. Wei, T.-T. Kao, M. M. Satter, S.-C. Shen, P. Douglas

Yoder, T. Detchprohm, R. D. Dupuis, A. M. Fischer, and F. A. Ponce, “Growth

of high-quality AlN layers on sapphire substrates at relatively low temperatures by

metalorganic chemical vapor deposition,” Physica Status Solidi (b), vol. 252, no. 5,

pp. 1089–1095, 2015.

[120] Y. L. Tsai, C. L. Wang, P. H. Lin, W. T. Liao, and J. R. Gong, “Observation of com-

positional pulling phenomenon in AlxGa1−xN (0.4<x<1.0) films grown on (0001)

sapphire substrates,” Applied Physics Letters, vol. 82, no. 1, pp. 31–33, 2003.

[121] B. Liu, R. Zhang, J. G. Zheng, X. L. Ji, D. Y. Fu, Z. L. Xie, D. J. Chen, P. Chen,

R. L. Jiang, and Y. D. Zheng, “Composition pulling effect and strain relief mecha-

nism in AlGaN/AlN distributed Bragg reflectors,” Applied Physics Letters, vol. 98,

p. 261916, 2011.

[122] Z. Chen, Y. Pei, S. Newman, D. Brown, R. Chung, S. Keller, S. P. DenBaars,

S. Nakamura, and U. K. Mishra, “Growth of AlGaN/GaN/AlGaN double hetero-

junction field-effect transistors and the observation of a compositional pulling ef-

fect,” Applied Physics Letters, vol. 94, p. 171117, 2009.

[123] G.B.Stringfellow, “Compositional ordering in semiconductor alloys,” Mater. Res.

Soc. Symp. Proc., vol. 312, pp. 35–46, 1993.

[124] J. Tersoff, “Stress-driven alloy decomposition during step-flow growth,” Physical

Review Letters, vol. 77, no. 10, pp. 2017–2020, 1996.

146

[125] P. Venezuela, J. Tersoff, J. Floro, E. Chason, D. Follstaedt, F. Liu, and M. Lagally,

“Self-organized growth of alloy superlattices,” Letters to nature, vol. 397, pp. 678–

681, 1999.

[126] B. Reuters, M. Finken, A. Wille, B. Hollander, M. Heuken, H. Kalisch, and A. Ves-

can, “Relaxation and critical strain for maximum In incorporation in AlInGaN on

GaN grown by metal organic vapour phase epitaxy,” Journal of Applied Physics,

vol. 112, p. 093524, 2012.

[127] P. Bogusławski, K. Rapcewicz, and J. Bernholc, “Surface segregation and interface

stability of AlN/GaN , GaN/InN , and AlN/InN {0001} epitaxial systems,” Physical

Review B, vol. 61, no. 16, pp. 820–826, 2000.

[128] J. H. van der Merwe, “Crystal interfaces. Part II. Finite overgrowth,” Journal of

Applied Physics, vol. 34, pp. 123–127, 1962.

[129] J. H. Van Der Merwe and W. A. Jesser, “An exactly solvable model for calculating

critical misfit and thickness in epitaxial superlattices: Layers of equal elastic con-

stants and thicknesses,” Journal of Applied Physics, vol. 63, no. 5, pp. 1509–1517,

1988.

[130] J. Matthews and A. Blakeslee, “Defects in epitaxial multilayers I. Misfit disloca-

tions,” Journal of Crystal Growth, vol. 27, pp. 118–125, 1974.

[131] Hutchinson, J. W. and Z., Suo, “Mixed mode cracking in layered materials,” Ad-

vances in Applied Mechanics, vol. 29, pp. 64–191, 1992.

[132] Z. Wu, K. Nonaka, Y. Kawai, T. Asai, F. a. Ponce, C. Chen, M. Iwaya, S. Kamiyama,

H. Amano, and I. Akasaki, “Strain Relaxation Mechanisms in AlGaN Epitaxy on

AlN Templates,” Applied Physics Express, vol. 3, p. 111003, 2010.

147

[133] Z. Ren, Q. Sun, S.-Y. Kwon, J. Han, K. Davitt, Y. K. Song, a. V. Nurmikko, H.-K.

Cho, W. Liu, J. a. Smart, and L. J. Schowalter, “Heteroepitaxy of AlGaN on bulk

AlN substrates for deep ultraviolet light emitting diodes,” Applied Physics Letters,

vol. 91, p. 051116, 2007.

[134] J. Tersoff and F. K. LeGoues, “Competing relaxation mechanisms in strained layers,”

Physical Review Letters, vol. 72, no. 22, pp. 3570–3574, 1994.

[135] B. N. Pantha, R. Dahal, M. L. Nakarmi, N. Nepal, J. Li, J. Y. Lin, H. X. Jiang,

Q. S. Paduano, and D. Weyburne, “Correlation between optoelectronic and struc-

tural properties and epilayer thickness of AlN,” Applied Physics Letters, vol. 90,

p. 241101, 2007.

[136] X. Li, S. Sundaram, Y. El Gmili, F. Genty, S. Bouchoule, G. Patriache, P. Disseix,

F. Reveret, J. Leymarie, J.-P. Salvestrini, R. Dupuis, P. Voss, and A. Ougazzaden,

“MOVPE grown periodic AlN/BAlN heterostructure with high boron content,” Jour-

nal of Crystal Growth, vol. 414, pp. 119–122, 2015.

[137] J. Kim, Z. Lochner, M.-H. Ji, S. Choi, H. J. Kim, J. S. Kim, R. D. Dupuis, A. M.

Fischer, R. Juday, Y. Huang, T. Li, J. Y. Huang, F. a. Ponce, and J.-H. Ryou, “Origins

of unintentional incorporation of gallium in InAlN layers during epitaxial growth,

part II: Effects of underlying layers and growth chamber conditions,” Journal of

Crystal Growth, vol. 388, pp. 143–149, 2014.

[138] X. Li, S. Sundaram, Y. E. Gmili, T. Moudakir, F. Genty, O. Legrani, R. D. Dupuis,

P. L. Voss, J. P. Salvestrini, and A. Ougazzaden, “BAlN thin layers for deep UV

applications,” Physica Status Solidi (a), vol. 212, no. 4, pp. 745–750, 2015.

[139] S. Chuang and C. Chang, “K·P Method for strained wurtzite semiconductors,” Phys-

ical Review B, vol. 54, no. 4, pp. 2491–2504, 1996.

148

[140] J. E. Northrup, C. L. Chua, Z. Yang, T. Wunderer, M. Kneissl, N. M. Johnson, and

T. Kolbe, “Effect of strain and barrier composition on the polarization of light emis-

sion from AlGaN/AlN quantum wells,” Applied Physics Letters, vol. 100, p. 021101,

2012.

[141] S. Ivanov, D. Nechaev, A. Sitnikova, V. Ratnikov, M. Yagovkina, N. Rzheutskii,

E. Lutsenko, and V. Jmerik, “Plasma-assisted molecular beam epitaxy of Al(Ga)N

layers and quantum well structures for optically pumped mid-UV lasers on c-Al2O3,”

Semiconductor Science and Technology, vol. 29, p. 084008, 2014.

[142] H. Murotani, Y. Yamada, H. Miyake, and K. Hiramatsu, “Silicon concentration de-

pendence of optical polarization in AlGaN epitaxial layers,” Applied Physics Letters,

vol. 98, p. 021910, 2011.

[143] Z. Bryan, I. Bryan, S. Mita, J. Tweedie, Z. Sitar, and R. Collazo, “Strain dependence

on polarization properties of AlGaN and AlGaN-based ultraviolet lasers grown on

AlN substrates,” Applied Physics Letters, vol. 106, p. 232101, 2015.

[144] J. Baur, K. Maier, M. Kunzer, U. Kaufmann, and J. Schneider, “Determination of the

GaN/AlN band offset via the (-/0) acceptor level of iron,” Applied Physics Letters,

vol. 65, no. 17, pp. 2211–2213, 1994.

[145] M. B. Nardelli, K. Rapcewicz, and J. Bernholc, “Strain effects on the interface prop-

erties of nitride semiconductors,” Physical Revew B, vol. 55, no. 12, p. R7323, 1997.

[146] N. Binggeli, P. Ferrara, and A. Baldereschi, “Band-offset trends in nitride hetero-

junctions,” Physical Review B, vol. 63, p. 245306, 2001.

[147] F. Bernardini and V. Fiorentini, “Macroscopic polarization and band offsets at nitride

heterojunctions,” Physical Revew B, vol. 57, p. R9427, 1998.

149

[148] G. Martin, A. Botchkarev, A. Rockett, and H. Morkoc, “Valence-band discontinu-

ities of wurtzite GaN, AlN, and InN heterojunctions measured by x-ray photoemis-

sion spectroscopy,” Applied Physics Letters, vol. 68, no. 18, pp. 2541–2453, 1996.

[149] J. Hopfield, “Fine structure in the optical absorption edge of anisotropic crystals,”

Journal of Physics and Chemistry of Solids, vol. 15, pp. 97–107, 1960.

[150] X. Li, S. Sundaram, P. Disseix, G. Le Gac, S. Bouchoule, G. Patriarche, F. Reveret,

J. Leymarie, Y. E. Gmili, T. Moudakir, F. Genty, J.-P. Salvestrini, R. D. Dupuis, P. L.

Voss, and A. Ougazzaden, “AlGaN-based MQWs grown on a thick relaxed AlGaN

buffer on AlN templates emitting at 285 nm,” Optical Materials Express, vol. 5,

no. 2, pp. 380–392, 2015.

[151] M. Shiojiri, C. C. Chuo, J. T. Hsu, J. R. Yang, and H. Saijo, “Structure and formation

mechanism of V defects in multiple InGaN/GaN quantum well layers,” Journal of

Applied Physics, vol. 99, p. 073505, 2006.

[152] H.-L. Tsai, T.-Y. Wang, J.-R. Yang, C.-C. Chuo, J.-T. Hsu, Z.-C. Feng, and M. Sh-

iojiri, “Observation of V Defects in Multiple InGaN/GaN Quantum Well Layers,”

Materials Transactions, vol. 48, no. 5, pp. 894–898, 2007.

[153] L. C. Le, D. G. Zhao, D. S. Jiang, L. Li, L. L. Wu, P. Chen, Z. S. Liu, Z. C. Li,

Y. M. Fan, J. J. Zhu, H. Wang, S. M. Zhang, and H. Yang, “Carriers capturing of

V-defect and its effect on leakage current and electroluminescence in InGaN-based

light-emitting diodes,” Applied Physics Letters, vol. 101, p. 252110, 2012.

[154] F. Lin, N. Xiang, P. Chen, S. Y. Chow, and S. J. Chua, “Investigation of the V-pit

related morphological and optical properties of InGaNGaN multiple quantum wells,”

Journal of Applied Physics, vol. 103, p. 043508, 2008.

[155] A. M. Armstrong, M. H. Crawford, and D. D. Koleske, “Contribution of deep-level

150

defects to decreasing radiative efficiency of InGaN/GaN quantum wells with increas-

ing emission wavelength,” Applied Physics Express, vol. 7, p. 032101, 2014.

[156] X. A. Cao, K. Topol, F. Shahedipour-Sandvik, J. Teetsov, P. M. Sandvik, S. E.

LeBoeuf, A. Ebong, J. W. Kretchmer, E. B. Stokes, S. Arthur, A. E. Kaloyeros,

and D. Walker, “Influence of defects on electrical and optical characteristics of

GaN/InGaN-based light-emitting diodes,” Solid State Lighting II, Proceedings of

SPIE, vol. 4776, pp. 105–113, 2002.

[157] S.-R. Jeon, S.-J. Lee, S. H. Jung, S. H. Lee, J. H. Baek, H. Jeong, O. H. Cha, E.-K.

Suh, and M. S. Jeong, “Effect of V-shaped defects on structural and optical properties

of AlGaN/InGaN multiple quantum wells,” Journal of Physics D: Applied Physics,

vol. 41, p. 132006, 2008.

[158] M. Moseley, A. Allerman, M. Crawford, J. J. Wierer, M. Smith, and L. Bieder-

mann, “Electrical current leakage and open-core threading dislocations in AlGaN-

based deep ultraviolet light-emitting diodes,” Journal of Applied Physics, vol. 116,

p. 053104, 2014.

[159] M. Shatalov, W. Sun, R. Jain, A. Lunev, X. Hu, A. Dobrinsky, Y. Bilenko, J. Yang,

G. a. Garrett, L. E. Rodak, M. Wraback, M. Shur, and R. Gaska, “High power AlGaN

ultraviolet light emitters,” Semiconductor Science and Technology, vol. 29, no. 8,

p. 084007, 2014.

[160] N. Kamata, A. Z. M. Touhidul Islam, M. Julkarnain, N. Murakoshi, T. Fukuda, and

H. Hirayama, “Nonradiative centers in deep-UV AlGaN-based quantum wells re-

vealed by two-wavelength excited photoluminescence,” Physica Status Solidi (b),

vol. 252, no. 5, pp. 936–939, 2015.

[161] X. Li, G. Le Gac, S. Bouchoule, Y. El Gmili, G. Patriarche, S. Sundaram, P. Disseix,

F. Reveret, J. Leymarie, J. Streque, F. Genty, J.-P. Salvestrini, R. Dupuis, X.-H. Li,

151

P. Voss, and A. Ougazzaden, “Structural and optical investigations of AlGaN MQWs

grown on a relaxed AlGaN buffer on AlN templates for emission at 280 nm,” Journal

of Crystal Growth, vol. 432, pp. 37–44, 2015.

[162] Z. Liliental-Weber, “Derivation of growth mechanism of nano-defects in GaN from

TEM data,” Journal of Electron Microscopy, vol. 49, no. 2, pp. 339–348, 2000.

[163] Z. Liliental-Weber, Y. Chen, S. Ruvimov, and J. Washburn, “Formation mechanism

of nanotubes in GaN,” Physcal Review Letters, vol. 79, no. 15, pp. 2835–2838, 1997.

[164] M. Leroux, N. Grandjean, B. Beaumont, G. Nataf, F. Semond, J. Massies, and

P. Gibart, “Temperature quenching of photoluminescence intensities in undoped and

doped GaN,” Journal of Applied Physics, vol. 86, pp. 3721–3728, 1999.

[165] J. Bai, Q. Wang, and T. Wang, “Greatly enhanced performance of InGaN/GaN

nanorod light emitting diodes,” Physica Status Solidi (a), vol. 209, no. 3, pp. 477–

480, 2012.

[166] P. Dong, J. Yan, Y. Zhang, J. Wang, C. Geng, H. Zheng, X. Wei, Q. Yan, and J. Li,

“Optical properties of nanopillar AlGaN/GaN MQWs for ultraviolet light-emitting

diodes,” Optics Express, vol. 22, no. S2, pp. A320–A327, 2014.

[167] M. A. Reshchikov and H. Morko, “Luminescence properties of defects in GaN,”

Journal of Applied Physics, vol. 97, p. 061301, 2005.

[168] A. Y. Polyakov and I.-H. Lee, “Deep traps in GaN-based structures as affecting

the performance of GaN devices,” Materials Science and Engineering: R: Reports,

vol. 94, pp. 1–56, 2015.

[169] M. Kneissl, T. Kolbe, C. Chua, V. Kueller, N. Lobo, J. Stellmach, a. Knauer, H. Ro-

driguez, S. Einfeldt, Z. Yang, N. M. Johnson, and M. Weyers, “Advances in group

152

III-nitride-based deep UV light-emitting diode technology,” Semiconductor Science

and Technology, vol. 26, p. 014036, 2010.

[170] A. V. Tikhonravov, M. K. Trubetskov, A. A. Tikhonravov, and A. Duparre, “Effects

of interface roughness on the spectral properties of thin films and multilayers.,” Ap-

plied Optics, vol. 42, no. 25, pp. 5140–5148, 2003.

[171] A. Hafaiedh and N. Bouarissa, “Energy gaps and optical properties for the quater-

nary AlxGayIn1−x−yN matched to GaN substrate,” Materials Chemistry and Physics,

vol. 115, no. 1, pp. 122–125, 2009.

[172] P. Herve and L. Vandamme, “General relation between refractive index and energy

gap in semiconductors,” Infrared Physics & Technology, vol. 35, no. 4, pp. 609–615,

1994.

[173] S. Krishnankutty, R. M. Kolbas, M. A. Khan, J. N. Kuznia, J. M. V. Hove, and D. T.

Olson, “Photoluminescence Characterization of AIGaN-GaN Pseudomorphic Quan-

tum Wells and Calculation of Strain Induced Bandgap Shifts,” Journal of Electronic

Materials, vol. 21, no. 6, pp. 5–8, 1992.

[174] “http://www.ioffe.ru/SVA/NSM/Semicond/BN/mechanic.html.”

[175] “http://accuratus.com/alumni.html.”

[176] R. M. Wentzcovitch, K. J. Chang, and M. L. Cohen, “Electronic and structural prop-

erties of BN and BP,” Physical Review B, vol. 34, pp. 1071–1079, 1986.

[177] H. Akamaru, A. Onodera, T. Endo, and O. Mishima, “Pressure dependence of

the optical-absorption edge of AlN and graphite-type BN,” Journal of Physics and

Chemistry of Solids, vol. 63, no. 5, pp. 887–894, 2002.

[178] P. Perlin, I. Gorczyca, N. E. Christensen, I. Grzegory, H. Teisseyre, and T. Suski,

153

“Pressure studies of gallium nitride: Crystal growth and fundamental electronic

properties,” Physical Review B, vol. 45, no. 23, pp. 13307–13313, 1992.

[179] S. Kim, I. P. Herman, J. A. Tuchman, K. Doverspike, L. B. Rowland, and D. K.

Gaskill, “Photoluminescence from wurtzite GaN under hydrostatic pressure,” Ap-

plied Physics Letters, vol. 67, no. 3, pp. 380–382, 1995.

[180] V. Tasco, I. Tarantini, A. Campa, A. Massaro, T. Stomeo, G. Epifani, A. Passaseo,

M. Braccini, M. C. Larciprete, C. Sibilia, and F. a. Bovino, “GaN/AlGaN microcav-

ities for enhancement of non linear optical effects,” in Photonic Crystal Materials

and Devices IX, Proc. of SPIE, vol. 7713, p. 77131Q, 2011.

[181] H. Hirayama, Y. Enomoto, A. Kinoshita, A. Hirata, and Y. Aoyagi, “Room-

temperature intense 320 nm band ultraviolet emission from quaternary InAlGaN-

based multiple-quantum wells,” Applied Physics Letters, vol. 80, no. 9, pp. 1589–

1591, 2002.

[182] H. Hirayama, S. Fujikawa, and N. Kamata, “Recent Progress in AlGaN-Based Deep-

UV LEDs,” Electronics and Communications in Japan, vol. 98, no. 5, pp. 1–8, 2015.

[183] D. B. Li, M. Aoki, T. Katsuno, H. Miyake, K. Hiramatsu, and T. Shibata, “Influ-

ence of growth interruption and Si doping on the structural and optical properties of

AlxGaN/AlN (x>0.5) multiple quantum wells,” Journal of Crystal Growth, vol. 298,

pp. 500–503, 2007.

154